Radiocarbon 2003

8 "*29",93557

' : ;; ;(,*"! ( 0<4# =0,M+

#" !

!$%

& ' ( ) *+,+- . / 0 ) 12+,34,35,! !!6557741333

! " ## !$#!" % "&' %"("$

' ) )* ( + * ,

' ) * + * , - -&./01231241"5 ' 2447)* , ) + * , 5 + 5

824479%:1;45448 9-:;45448 + 95< =5 - - ) )> - )> + 5 + ) ? - @@@5 ) 5 A 5 5

@ ) @ * @ 8 = - @& - 9 ) 5 @ ) 3 +* )* ) ) * 5

" O P @@@5 ) 5 A A ) 5 ) # - -( - + * , -D010< @ - -&./01231241"5 5%E1/24..134./0F<=%E1/24..134//DF % G ) 5 5

5 ! + ) ) *- + ) @@@5 ) 5 A A ) 5 5> ) * + ) * 3 -@ = )- 5 - ) 5 + #" ) ) )* *- @ ) )* 5

)% %AA@@@5 ) 5 A

= A ) )* @ % !" " #$ " # ! % # 8 9F #" & " & # !" & " 8"9F<"8 R " ? ? M"9F% " % " '" !" !5 Vo l 45 , N r 1 Radiocarbon 2003

CONTENTS

EDITORIAL BOARD ...... iii

FROM THE EDITOR ...... v

FROM THE FORMER MANAGING EDITOR ...... vii

ARTICLES Investigation of a Chinese Ink Rubbing by 14C AMS Analysis Hong-Chien Yuan, Walter Kutschera, Tze-Yue Lin, Peter Steier, Christof Vockenhuber, Eva Maria Wild ...... 1 Radiocarbon Age Offsets Between Living Organisms from the Marine and Continental Reservoir in Coastal Localities of Patagonia (Argentina) Roberto R Cordero, Héctor Panarello, Sonia Lanzelotti, Cristian M Favier Dubois ...... 9 Radiocarbon Dating of Individual Fatty Acids as a Tool for Refining Antarctic Margin Sediment Chronologies Naohiko Ohkouchi, Timothy I Eglinton, John M Hayes ...... 17 14C Chronology of Late Pleistocene–Holocene Events in the Nizhnee Priamurie (Southeast Russia) V B Bazarova, L M Mokhova, L A Orlova, M A Klimin, I G Gvozdeva ...... 25

Archaeology On the Coexistence of Man and Extinct Pleistocene Megafauna at Gruta del Indio (Argentina) Alejandro García ...... 33 El Mirón Cave and the 14C Chronology of Cantabrian Spain Lawrence Guy Straus, Manuel González Morales ...... 41 14C Absolute Chronology of Pyramid III and the Dynastic Model at Pachacamac, Peru Adam MichczyÒski, Peter Eeckhout, Anna Pazdur ...... 59

Calibration High-Precision AMS 14C Results on TIRI/FIRI Turbidite Thomas P Guilderson, John R Southon, Thomas A Brown ...... 75 Radiocarbon Calibration for Japanese Wood Samples Minoru Sakamoto, Mineo Imamura, Johannes van der Plicht, Takumi Mitsutani, Makoto Sahara ...... 81

Corals and Shells Decadal Timescale Shift in the 14C Record of a Central Equatorial Pacific Coral A G Grottoli, S T Gille, E R M Druffel, R D B Dunbar ...... 91

Soils and Sediments Dating of Prehistoric Burial Mounds by 14C Analysis of Soil Organic Matter Fractions Søren M Kristiansen, Kristian Dalsgaard, Mads K Holst, Bent Aaby, Jan Heinemeier ...... 101

i NOTES & COMMENTS Balanced Window Method in 14C Liquid Scintillation Counting P Theodórsson, S Ingvarsdottir, G I Gudjonsson ...... 113

DATE LIST Radiocarbon Dates from Halfiah Gibli (Abadiyeh), a Predynastic Settlement in Upper Egypt Kathryn A Bard...... 123

RADIOCARBON UPDATES ...... 131

ERRATUM ...... 133

ii FROM THE EDITOR

In this issue, we now have a new managing editor, Mark McClure, and Agnieszka Baier continues as business manager. Kim Elliott has left the university. The cause of her departure is a beautiful daughter, Corinne. I am sure everyone wishes Kim well. Kim writes her own farewell after this editorial. It is difficult to start these comments without noting the passing of John Head this summer 2003. John’s work in the dating of sediments and studies of charcoal and cellulose are well-known. We will all remember John’s quiet demeanor and dedication to his work. Weijian Zhou, who worked closely with John for many years in Xi’an and Australia, will write an obituary for our next issue. The obituary was also presented at the 18th International Radiocarbon Conference. In this issue, we again present a selection of interesting papers in the field of radiocarbon dating. A paper by Yuan et al. discusses dating of Chinese papers and Okhouchi et al. delve into the mysteries of trying to separate (and date) fatty acids from sediments. Reservoir effects, both applied to dating (Cordero et al.) and changes in the 14C in corals (Grottoli et al.), are discussed. Papers by Bazarova et al. and Garcia are concerned with Late Pleistocene megafauna and the interaction of early man with these species. Straus and González Morales are similarly interested in the dating of early modern humans. Our remaining papers discuss archaeological issues, in Peru (MichzyÒski et al.) and prehistoric burial mounds (Kristiansen et al.), as well as a date list of an Egyptian site (Bard). Finally, we have some technical discussions on calibration for wood samples (Sakamoto et al.), the precision of AMS (Guilderson et al.), and the methodology of scintillation counting (Theodórsson et al.). I hope that this variety of topics is of interest; our next issue will focus on the Fourth International Radiocarbon Intercomparison (FIRI). I enjoyed meeting many new and old friends at the recent Radiocarbon Conference and we also look forward to the Proceedings of that conference in 2004. A J Timothy Jull

v FROM THE FORMER MANAGING EDITOR

Kimberley Tanner Elliott with daughter, Corinne, born in March 2003

On March 14 of this year, my husband Garnett and I became parents for the first time. The arrival of our daughter Corinne Marie marked the end of one career (at least for the time being) and the beginning of another. However, the journal is now in the very capable hands of Mark McClure and Agnieszka Baier. Many of you will get to meet them at the Radiocarbon Conference in New Zealand. My career at Radiocarbon spanned nearly 10 years and I am grateful to have worked with so many fascinating people from around the world during that time. Few careers provide such an opportunity. I was hired by Renee Kra in the fall of 1993 as assistant editor. In the beginning, my primary responsibilities were typesetting, proofing, and managing orders and subscriptions. David Sewell and I began working for the journal the same week. Renee’s energy and dedication were inspiring, and with David’s help as senior assistant editor, we modernized the journal’s equipment to create a one-stop publishing house. We acquired desktop- publishing software and graphics scanning equipment and soon after, David produced our first webpage. The only thing we lacked was the printing press.

vii viii From the Former Managing Editor

In the summer of 1994, I attended my first Radiocarbon Conference, which was held in Glasgow. I was pleased to work closely with the gracious organizers (Gordon Cook, Doug Harkness, Angus Mackenzie, Brian Miller, and Marian Scott). That was my first opportunity to meet the faces behind the names on all those manuscripts I’d been typesetting and mailing off. The radiocarbon community is diverse but I found everyone to be friendly and approachable. I will miss seeing so many old friends at future radiocarbon meetings. The journal faced financial trouble in 1994 as subscribing libraries began tightening their belts, and we lost a substantial number of customers. So, a few months after the Glasgow meeting, Renee was forced to lay me off and run the journal on a skeleton crew of just herself, David, and a part-time student assistant. (I found work at another university department.) Fortunately, I was able to return in the summer of 1995 after a new budget was implemented. Nineteen ninety-five was a busy year, as Renee, David, and I worked on three conference proceedings books at once, Tree Rings, Environment and Humanity (edited by Jeffrey S Dean, David M Meko, and Thomas W Swetnam), LSC 94 (edited by Gordon T Cook, Douglas D Harkness, Angus B Mackenzie, Brian F Miller, and E Marian Scott), and the Glasgow Radiocarbon Conference (Vol. 37, Nr. 2, 1995). In 1996, we were fortunate to bring on board Tim Jull as consulting editor. Tim was later appointed editor upon the retirement of Austin Long in 1999. Tim was joined then by Warren Beck and George Burr as associate editors (Vol. 41, Nr. 3, 1999). In 1997, David Sewell and I attended the Radiocarbon Conference in Groningen, the Netherlands. Sadly, Renee was too ill to make the trip and had to step back from the journal to look after her health. David and I prepared the Groningen proceedings issues (Vol. 40, Nr. 1–2, 1998) with the dedicated cooperation of conference organizers Wim Mook and Hans van der Plicht. Because there were so many papers, those proceedings were the first to be published in two books. Because of her medical problems, Renee retired earlier than we all expected, and in 1997 David offi- cially assumed the role of managing editor and I was promoted to senior assistant editor, and later assistant managing editor. When David relocated to Virginia in 1999, I took over as managing editor. In 2000, one of my first projects was editing the special Radiocarbon issue in honor of Renee Kra’s three decades of dedication to the journal (Vol. 42, Nr. 1, 2000). Also in 2000, Radiocarbon partnered with Catchword/Ingenta to publish the journal online. The same year, I attended my third Radiocarbon Conference, in the Judean Hills of Israel. Conference organizers Israel Carmi, Elisabetta Boaretto, and Hendrik Bruins worked tirelessly with me to publish the mammoth proceedings issues (three books this time!). Assistant Editor Agnieszka Baier joined me in 2001 and put in many hours on Vol. 43, Nr. 1–3, 2001. I also had the pleasure of working with Hans van der Plicht again in 1998 on the IntCal98 calibration issue, Vol. 40, Nr. 3, 1998 (he designed the Dali-esque cover art, by the way, but never got proper credit. Sorry, Hans!). Other recent projects have included a special issue with Yaroslav Kuzmin in 2002 (“Old and New World Connections,” Vol. 44, Nr. 2, 2002) and the Varve/Comparison issue, Vol. 42, Nr. 3. The folks at Radiocarbon have been kind enough to let me keep my email address [email protected], so I hope to hear from some of my old friends from time to time. Any email regarding Radiocarbon business should go to Mark McClure at [email protected].

INVESTIGATION OF A CHINESE INK RUBBING BY 14C AMS ANALYSIS

Hong-Chien Yuan1 • Walter Kutschera2,3 • Tze-Yue Lin4 • Peter Steier2 • Christof Vockenhuber2• Eva Maria Wild2

ABSTRACT. The date of a Chinese ink rubbing was determined using radiocarbon accelerator mass spectrometry (AMS) to be in the range from AD 1480 to AD 1670 (95.4% confidence limit). Together with a scanning electron miscroscope (SEM) analysis of the ink and a comparative study of the Chinese characters, it was determined that the ink rubbing must have been performed before Emperor Kang Hsi (AD 1662–1722), who ruled at the beginning of the Chin Dynasty. On the other hand, the stone stele, from which the ink rubbing was produced, was carved in AD 531, which is consistent with an analysis of some erased characters. Such analysis seems to be useful to help clarify possible forgeries of these art objects.

INTRODUCTION Ink rubbing has been an important technique in Chinese culture. Before the development of modern printing processes with colortype, ink rubbing was a unique tool available to faithfully reproduce original artistic work, in particular, calligraphy. In reality, this technique was closely linked to the progress of various types of Chinese writing, including seal-, official-, running-, regular-, and cur- sive-script. While the engraving of writing on a stele (engraved stone) bears the significance of conservation of art on perishable paper or silk, ink rubbing favors the access of calligraphy for wider circulation. It has sustained the advance of calligraphy for more than 2000 yr, and prolonged the popularity of invaluable master works, accordingly. Ink rubbing is similar to the letterpress, for which the original work is carefully duplicated by writing on a polished stone surface, and then, the characters are engraved with a chisel to form a concave inscription. For reproduction, a piece of pre-wetted soft paper is first pressed unto the incised stone to shape the characters. After drying, the paper is pounded with an ink-soaked cotton pad to show the engraved characters in white. In another way, the paper is treated by pre-wetting and pressing as in the first one. After peeling off and drying, a thin layer of Chinese ink of suitable consistency is applied over the stone surface, followed by careful matching the said paper with the inked surface below. A sheet of waxed blanket is placed above. By rubbing the blanket downward with a pad to permeate ink through the paper, the blank characters appear amidst the inked background. Although the time of preparing the stele may have been recorded, the date of rubbing was not to be written down. This practice has become a chance for forgery. As an art object, ink rubbings have been highly appreciated by scholars and aristocrats. Forgery of master works of ink rubbings has become a serious problem in art conservation. The authenticity of an article can be challenged by a counterfeit of newer ones pretending to be old rubbings, or by imitation of original stele through new stone engravings. In this respect, 14C dating by the AMS technique is a viable approach in determining the time period when the rubbing was performed. The purpose of this investigation is to apply 14C dating to an ink rubbing from an inscribed gravestone of Chang Hsuan (AD 461–493) from the Northern Wei Dynasty. The stele was laid in AD 531 in the Yunchi area, Shanxi Province, but neither the date when the stele was unearthed nor the date of later disappearance were recorded. In 1825, a calligrapher, Ho Shao Chi, procured the ink rubbing

1Donaufelder Strasse 246/4, A-1220 Vienna, Austria. 2Vienna Environmental Research Accelerator (VERA), Institut für Isotopenforschung und Kernphysik, Universität Wien, Währinger Strasse 17, A-1090 Vienna, Austria. 3Corresponding author. Email: [email protected]. 4Industrial Technology Research Institute, Hsinchu, Taiwan, China.

1 2 H-C Yuan et al. of this stele from a book market in Jinan, Shantung Province, and considered it to be the sole article in existence. This ink rubbing, also named the stele of Chang Hei Nue (to be explained later), is now part of the Shanghai Museum collection. However, its context and calligraphy differ from those of the author’s (H-C Yuan’s) family collection. Several forged errors can be obviously detected in the evaluation of the content (Yuan 2003). Thus, a sample was taken from the Yuan family collection for 14C dating by AMS at the VERA facility to estimate the time period when this rubbing was performed. The nature of the ink used for the rubbing was examined by a scanning electron microscope (SEM) to assist the interpretation of the results.

MATERIALS Chinese ink used in rubbing is a fine dispersion of carbon black and glue from bone collagen in the form of a viscous paste with suitable consistency. Carbon black can be either from pine soot or as lamp black from Tung oil, the latter being from annual seed production. Pine soot, however, is produced from logged pines with ages ranging from decades to hundreds of years. In addition, the blending of old, well-aged inks into new preparation (probably for viscosity and particle size adjust- ment) had been a trade secret. Thus, if the dated paper is contaminated with a large amount of ink, a shift toward an older age cannot be excluded. But, although the exact amount of carbon added this way to the paper sample could not be determined, it is estimated to be well below 10% of the total carbon from the paper being dated, resulting in a negligible shift of the age. In contrast, the raw material for Chinese paper were fibers from Broussonetia trees or bamboo, both being fast-growing species. Thus, the problem of old paper material discussed in Burleigh and Baynes-Cope (1983) can be excluded.

METHODS

Sample Preparation A strip (1 × 26.7 cm) was cut from the edge of the ink rubbing paper without touching the character part (Figure 1). Approximately one-half of this sample was taken from the strip and processed further for the 14C age determination (VERA laboratory number V-2268). The material was first cleaned ultrasonically, followed by a standard Acid-Base-Acid (ABA) treatment at 60 °C using 1M HCl–0.1M NaOH–1M HCl in sequence, and rinsing with bi-distilled water between the change of the reagent. Ten mg of the pretreated and dried material was transferred into a quartz tube together with 1 g CuO and some silver wire, then evacuated and sealed with a torch. The sample was combusted in a muffle furnace for 4 hr at a temperature of 900 °C. The evolved CO2 was reduced to carbon with hydrogen in the presence of an Fe catalyst at 580 °C. The resulting graphite-catalyst mixture was pressed into an Al target holder for AMS analysis. For reference, targets were prepared from the IAEA standards C-3 (cellulose), C-5 (subfossil wood) and C-6 (ANU-sucrose), while dead carbon from a graphite rod was used as machine and chemistry blanks (Wild et al. 1998). As far as possible, the standards and the blank material was treated in the same way as the sample. The 14C/12C ratio of the chemistry blank was subtracted from the 14C/12C ratios of both the standards and the sample for background correction.

AMS Measurements The VERA AMS system is based on a 3-MV Pelletron tandem accelerator (Kutschera et al. l997; Priller et al. 1997). Procedures for 14C and δ13C measurements have been described previously (e.g., Rom et al. 1998). The 14C measurements are now performed in a fully automated way (Steier et al. AMS Investigation of Chinese Ink Rubbing 3

Figure 1 Sample taken from the edge of the ink rubbing (see text for details)

2000; Puchegger et al. 2000). Recently, VERA has been upgraded to accommodate isotope ratio measurements across the nuclear chart (Vockenhuber et al. 2003). This now allows one to switch quickly between high-precision “routine” 14C measurements (e.g., Wild et al. 2001) and more elab- orate AMS measurements up to 244Pu (Steier et al. 2003; Winkler et al. 2003). In the present investigation, the accelerator was operated at a terminal voltage of 2.7 MV. Typically, 12C currents from the MC-SNICS cesium beam sputter source were in the range of 28 to 42 µA. For fast switching between carbon isotope measurements, the beam-sequencing method was used (Priller et al. 1997), with sequential time periods for isotope injection of 213 ms for 14C, 1.15 ms for 13C, and 0.15 ms for 12C. The sample target was measured 6 times with each measuring period lasting about 5 minutes.

Age Determinations The 14C content in percent Modern Carbon (pMC) and the 14C ages of the ink rubbing sample were calculated according to Stuiver and Polach (1977) from the 14C/12C and the 13C/12C (δ13C) ratios, both measured at VERA. The 14C age of the rubbing sample was converted by the OxCal software version 3.5 (Bronk Ramsey 2000, 2001) with atmospheric calibration data from Stuiver et al. (1998) into a calibrated age.

Scanning Electron Microscope (SEM) Investigations A tiny fragment of the ink rubbing paper was gold-plated for examination by SEM, using a 15 kV Norton S-4200 SEM. For identifying the origin of carbon in ink, photos were taken at 10k, 20k, and 50k magnifications with operating voltage of 10 kV. These photomicrographs were compared with those from paper bearing inks from either pine soot or lamp black. 4 H-C Yuan et al.

RESULTS AND DISCUSSION The result of the OxCal calibration of the measured 14C age is shown in Figure 2 and the results are summarized in Table 1.

Atmospheric data from Stuiver et al. (1998); OxCal v3.5 Bronk Ramsey (2000); cub r:4 sd:12 prob usp[chron]

600BP VERA-2268 : 295±35BP 68.2% probability 500BP 1520AD (48.8%) 1590AD 1620AD (19.4%) 1660AD 400BP 95.4% probability 1480AD (95.4%) 1670AD 300BP

200BP

100BP Radiocarbon determination

0BP

1400CalAD 1600CalAD 1800CalAD 2000CalAD Calibrated date Figure 2 Plot of 14C age versus calibrated date for Sample V-2268 determined with the Oxcal calibration program

Table 1 13C and 14C results of the ink rubbing sample Laboratory δ13Ca 14C contenta 14C agea number (‰) (pMC) (yr BP) Calibrated dateb VERA-2268 –26.2 ± 0.6 96.4 ± 0.4 295 ± 35 AD 1480 (95.4%) AD 1670 a1σ uncertainty bCalibrated range for the 95.4% confidence (2 σ) limit determined with the OxCal calibration program (see Figure 2)

Analysis of the results shows that the rubbing was performed before AD 1670, towards the end of the Ming Dynasty in China. This estimate is supported by a clue in the inscription. Although the name of the inscribed stele is in the memory of Chang Hsuan, in the Chin Dynasty it was also called the stele of Chang Hei Nue, as mentioned earlier. The reason was that the Emperor Kang Hsi in the Chin Dynasty had a given name of “Hsuan” Hua. The imperial regulations demanded that no character in the stele should coincide with the given name of any emperor in that dynasty, otherwise, it would become a serious offence. For any existing steles with character(s) in conflict with this regu- lation, such character(s) should be chiseled off, producing a blank in the corresponding spot of the ink rubbings. Since the given name “Hsuan” in the stele was intact as indicated in this rubbing (Fig- ure 3a), it is certain that the rubbing was done before the Kang Hsi epoch (AD 1662–1722), as con- AMS Investigation of Chinese Ink Rubbing 5 firmed by the AMS measurement results. The above-mentioned practice is supported by the same stele in which the character “Shun” was mostly chiseled off (yet still identifiable as this particular character), it coincided with the given name of Emperor Hsiung-tsung, Li Shun (AD 806–820), in the late Tang Dynasty (Figure 3c). Thus, the original gravestone was unearthed before the end of Tang Dynasty, AD 907. The given name of Chang Hsuan’s father was completely carved away in the stele (Figure 3b); therefore, it could not be determined whether this character coincides with that of the ruling monarch.

10 cm A C B

Figure 3 Part of the ink rubbing showing important locations of characters: A–“Hsian” is intact; B–father’s name unknown; C–“Shun” is chiseled away

Figures 4A–4C show the SEM photomicrographs of ink particles magnified to 10k, 20k, and 50k, respectively. By comparing the figures from the present ink rubbing with 4D (from pine soot) and 4E (from lamp black), the cloudy appearance of Figures 4A–4C and 4D clearly indicates that the ink used in this rubbing was based on pine soot. As discussed above, the influence of the ink on the date of the paper should be negligible.

CONCLUSION The use of AMS for dating the paper of Chinese ink rubbing has been successfully accomplished with good precision. Its result could be assessed with pertinent information of historical events. In the light of wide-spread forgery of classical Chinese paintings and calligraphy works, the use of 14C dating of paper and silk substrates with the AMS technique can help resolve authenticity problems 6 H-C Yuan et al.

Figure 4 Scanning electron microscope (SEM) photographs of ink particles from the investigated ink rubbing material (A–C), contemporary pine soot (D), and of lamp black from AD 1810 (E)

and aid screening art objects. For the majority of art works, samples can be taken from their edges without touching the colored part.

ACKNOWLEDGEMENT H-C Yuan would like to thank Mr Chiu Tso-Chie for his arrangement of SEM activities.

REFERENCES

Bronk Ramsey C. 2000. The OxCal Program version 3.5 25(2):669–74. URL: http://www.rlaha.ox.ac.uk/orau/index.htm. Kutschera W, Collon P, Friedmann H, Golser R, Hille P, Bronk Ramsey C. 2001. Development of the radiocarbon Priller A, Rom W, Steier P, Tagesen S, Wallner A, calibration program. Radiocarbon 43(2A):355–63. Wild E, Winkler G. 1997. VERA: a new AMS facility Burleigh R, Baynes-Cope AD. 1983. Possibilities in the in Vienna. Nuclear Instruments and Methods in Phys- dating of writing materials and textiles. Radiocarbon ics Research B 123:47–50. AMS Investigation of Chinese Ink Rubbing 7

Priller A, Golser R, Hille P, Kutschera W, Rom W, Steier ibration, 24,000–0 BP. Radiocarbon 40(3):1041–83. P, Walner A, Wild E. 1997. First performance tests of Vockenhuber C, Ahmad I, Golser R, Kutschera W, VERA. Nuclear Instruments and Methods in Physics Liechtenstein V, Priller A, Steier P, Winkler S. 2003. Research B 123:193–8. Accelerator mass spectrometry of heavy long-lived ra- Puchegger S, Rom W, Steier P. 2000. Automated evalua- dionuclides. International Journal of Mass Spectron- tion of 14C AMS measurements. Nuclear Instruments omy 223–224:713–32. and Methods in Physics Research B 172:274–80. Wild E, Golser R, Hille P, Kutschera W, Priller A, Pu- Rom W, Golser R, Kutschera W, Priller A, Steier P, Wild chegger S, Rom W, Steier P, Vycudilik W. 1998. First E. 1998. Systematic investigations of 14C measure- 14C results from archaeological and forensic studies at ments at the Vienna Environmental Research Acceler- the Vienna Environmental Research Accelerator. Ra- ator. Radiocarbon 40(1):255–63. diocarbon 40(1):273–81. Steier P, Puchegger S, Golser R, Kutschera W, Priller A, Wild EM, Stadler P, Bondár M, Draxler S, Friesinger H, Rom W, Wallner A, Wild E. 2000. Developments to- Kutschera W, Priller A, Rom W, Ruttkay E, Steier P. wards a fully-automated AMS system. Nuclear In- 2001. New chronological frame for the young struments and Methods in Physics Research B 161– Neolithic Baden culture in central Europe. Radiocar- 163:250–4. bon 43(2B):1057–63. Steier P, Golser R, Kutschera W, Priller A, Vockenhuber Winkler S, Ahmad I, Golser R, Kutschera W, Orlandini C, Winkler S. Forthcoming. VERA, an AMS facility KA, Paul M, Priller A, Steier P, Valenta A, Vockenhu- for “all” isotopes. Nuclear Instruments and Methods ber C. Forthcoming. Developing a detection method in Physics Research B. of environmental 244Pu. Nuclear Instruments and Stuiver M, Polach HA. 1977. Discussion: reporting of Methods in Physics Research B. 14C data. Radiocarbon 19(3):355–63. Yuan H-C. Forthcoming. On the authenticity of Chang Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, Hsuan Stele. Taiwan: National Museum Monthly of Hughen KA, Kromer B, McCormac G, van der Plicht Chinese Art. In Chinese. J, Spurk M. 1998. INTCAL 1998 radiocarbon age cal- RADIOCARBON, Vol 45, Nr 1, 2003, p 9–15 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

RADIOCARBON AGE OFFSETS BETWEEN LIVING ORGANISMS FROM THE MARINE AND CONTINENTAL RESERVOIR IN COASTAL LOCALITIES OF PATAGONIA (ARGENTINA)

Roberto R Cordero1,2 • Héctor Panarello1 • Sonia Lanzelotti3 • Cristian M Favier Dubois4

ABSTRACT. The radiocarbon of the local reservoir effect (RE) was observed in many sectors along the Argentinean Patag- onic coast. Results show variations in the 14C offsets and differences between marine and continental species growing within the same locality, ranging from about 80–1100 yr BP. It is postulated that such variations are mainly due to local factors, including the coast morphology and the contribution of continental waters. The relevance of these kinds of studies for the interpretation of age in archaeological samples is highlighted in this paper.

INTRODUCTION A conventional radiocarbon age is determined in relation to the concentration of 14C found in the sample analyzed and of the atmospheric carbon dioxide with which it is in equilibrium (atmospheric carbon reservoir) (Stuiver and Polach 1977; Stuiver 1978). The difference between the sample and “modern” activity makes it possible to assign an absolute age to the sample. However, the sample may not be in equilibrium with the atmospheric reservoir, but with other carbon reservoirs (seawater, rivers, lakes, estuaries, etc.) where the 14C concentration is different, originating what is called the reservoir effect (RE). These reservoirs are usually depleted in 14C, thus, they present older apparent ages. For instance, carbon inorganic species dissolved in seawaters globally present a 14C concentration of about 400 yr in samples in equilibrium with surface waters up to 80 m deep for the Northern Hemisphere (Bard et al. 1993; Stuiver and Brazionas 1993). This value has regional variations. The specific age determined has several possible causes: 1. Variations in the emergence of older, deep waters or “upwelling” (Robinson and Trimble 1981); 2. The contribution of continental waters in coastal regions; 3. The geometry of the basin (communication with the open sea); 4. The presence of rocks or fossils that contribute non-active 14C carbonates or carbonates with an activity different from the sea reservoir. All these processes affect the composition of the exoskeletons of some organisms (mollusks, crus- taceans, algae, coral, fishes, sea mammals, etc.) or the composition of sea sediments, mirroring the carbon isotopic relations of this reservoir. In coastal archaeological sites, remains from sea organisms are abundant and can show this phenomenon to a greater or lesser extent. When corrections by the RE are reported in 14C age records, they are estimated based on calibration programs because it is still not possible to consider local variations. This makes it necessary to focus all efforts towards a regional calibration (Albero et al. 1986, 1987; Figini 1999). Although this effect is correctly characterized in many coastal areas (e.g. Dye 1994; Ingram and Southon 1996), there is little information for the vast coast of Tierra del Fuego and Patagonia. For the Beagle Channel (Tierra del Fuego), there are estimations based on archaeological

1Instituto de Geocronología y Geología Isotópica (INGEIS-CONICET), Pabellón INGEIS, Ciudad Universitaria. (1428) Buenos Aires, Argentina. 2Corresponding author. Email: [email protected] or [email protected]. 3Facultad de Filosofía y Letras, Universidad de Buenos Aires. Puán 480 (1406) Buenos Aires, Argentina. Email: [email protected]. 4CONICET, INCUAPA, Departamento de Arqueología, Universidad Nacional del Centro de la Prov. de Buenos Aires, Av del Valle 5737 (7400) Olavarría, Argentina. Email: [email protected].

9 10 R R Cordero et al. sites of 620 ± 140 yr (Albero et al. 1986), in reference to an average between different sea species, and of 556 ± 61 yr (Albero et al. 1987) for Mytilus (mussel). The latter is used as a correction for 14C ages of sea organisms found in prehistoric settlements (middens) in the area. This contribution presents estimations for modern 14C age offsets obtained in 20 localities of the Fuego-Patagonic coast and 1 value from the Antarctic sector. Some advantages for calculating the 14C age offsets on modern samples are the possibility to assure the contemporaneity between them (often difficult in archaeological contexts) and omitting the problems originated by the postdepositional contamination of the sample (either pedogenic or diagenic), a factor that may introduce some complex variations. In this work, we do not attempt to calculate the modern RE. The RE formerly estimated at 40 yr for the Southern Hemisphere has shown significant differences in calculations made for southern Chile, New Zealand, and Tasmania by Stuiver et al (1998). This introduces an important ambiguity when calculating calibrated ages. Thus, here we present the differences in the 14C ages of species in equilibrium with the marine and atmospheric reservoirs belonging to the same localities (14C age offsets). With these data, it will be possible to estimate the local RE in the future. The calculation of the past local RE in relation to the modern local RE is valid if there were no significant climatic variations, either regional or local, that may have affected oceanic circulation patterns or coastal streams, among other factors. Palynological studies show minor environmental fluctuations during the last 4000 yr BP in the region (Markgraf 1993; Mancini 1998), when the Patagonian coast acquired its modern general shape after a prolonged sea transgression (Codignotto 1996). Accordingly, the oldest archeological site known in continental Patagonia has a date of 3220 ± 70 yr (LP-515), in the Península de Valdés (Gómez Otero 1994).

MATERIALS AND METHODS There were 21 pairs of modern samples collected. In each pair, 1 sample represents the equilibrium with atmospheric reservoir (vegetables) and the other sample with sea reservoir (mollusks). They come from different coastal localities in Patagonia and Tierra del Fuego, as well as 1 sample from Livingston Island (Antarctic) (See Figure 1). The 14C activity of the atmospheric reservoir sample is taken as the modern activity, corresponding to 0 yr BP (i.e. 1950). Using the ratio between the sea sample activity and the atmospheric one, an apparent age is assigned to the former. When the 14C concentration is normalized to modern atmospheric value, corrections due to the bomb effect (de Vries 1958) or the Suess effect (Suess 1955) are not significant. It is not that these effects are not present, but variations in study areas are much less important in magnitude that the effect studied. Most of the localities analyzed correspond to coastal sites with an archaeological interest. The ages were determined by the INGEIS laboratory (14C Analysis Laboratory, Buenos Aires, Argentina) according to the following established routines:

• Pretreatment of shells: washed with hot H2O2 200 vol., 2% of HCl, to remove the surface layer, with a loss in weight of 20%. • Pretreatment of vegetables: washed with hot (100 ºC) NaOH, 2%, to obtain a clean solution, then washed with HCl 5% (90 ºC). The sea samples studied correspond to the following meso- and infra-littoral mollusks: Mussel (Mytilus edulis), Little Mussel (Brachidontes spp.), Cholga Mussel (Aulacomya ater), and Limpet Age Offsets on the Coast of Patagonia (Argentina) 11

Figure 1 Coastal localities in Patagonia and Tierra del Fuego where dated samples were taken

(Patinigera sp.). They were collected in the high-tide undertow line, and appeared to be intact, bivalve species with their shells united by the ligament or, at least, with the periostracum well-preserved, so as to verify a recent death. Land samples correspond to short-life vegetal species (herbs) or young bushes’ sprouts from the same localities: Coirón (Festuca sp.), Calafate (Berberis spp.), and Mata verde (Lepidophyllum cupressiforme). In the case of the Antarctic sample, it was only possible to find leaf lichens. The ages assigned were obtained from only the sea species in each case, meaning that the interspe- cific variability and vital effects have not yet been considered (for the Beagle Channel, see Albero et al. 1986).

RESULTS AND DISCUSSION One of the initial suppositions of this work was the existence of a general increase in ∆R values with latitude, which is coherent with the oceanic pattern. This, however, was only suggested in open-sea localities with no influence from continental waters. As in other coastal environments, there is a very strong influence of local factors. In the Golfo San José, the ages obtained are very old. This may be due to the abundant presence of fossil carbonates in the basin, which, together with the restricted circulation (see Figure 1), may 12 R R Cordero et al.

Table 1 Species collected in the different localities and their 13C values Locality: Terrestrial Marine Genus/species δ13C Genus/species δ13CDepthDate [m] 1-Punta Norte Festuca sp. –22.0 Aulacomya ater 2.0 01/28/99 1.35 2-Golfo San José (Playa Festuca sp. –22.0 Limnoperna 2.0 01/28/99 1.45 Larralde), surficial water fortunie 3-Golfo San José (Playa Festuca sp. –22.0 Aulacomya ater 02/08/00 1.25 Larralde), deep water 4- Golfo Nuevo (Playa Festuca sp. –22.0 Aulacomya ater 2.3 02/08/00 1.20 Villarino) 5- Golfo Nuevo- Festuca sp. –22.0 Aulacomya ater 1.0 02/08/00 1.00 Puerto Madryn 6-Playa Unión Festuca sp. –22.0 Brachidontes 1.5 02/10/00 1.17 rodriguezi 7-Comodoro Rivadavia Festuca sp. –28.0 Aulacomya ater 1.9 02/10/00 1.26 8-Rada Tilly Festuca sp. –25.2 Aulacomya ater 1.0 02/10/00 1.35 9-Caleta Olivia Senecio sp. –28.0 Aulacomya ater –9.1 02/10/00 1.30 10-Isla Lobos Festuca sp. –25.0 Aulacomya ater 1.0 02/29/00 1.45 11-Bahía Desvelos Festuca sp. –25.0 Aulacomya ater 1.0 12/14/99 1.25 12-Bahía Laura Festuca sp. –25.0 Aulacomya ater –2.4 12/10/99 1.35 13-Puerto San Julián Poa sp. –20.9 Mytilus edulis 1.1 03/01/00 1.30 14-Punta Bustamante Lepidophyllum –25.8 Mytilus edulis 1.5 03/27/00 1.40 cupressiforme 15-Punta Loyola Lepidophyllum –23.7 Mytilus edulis 1.2 10/22/01 1.45 cupressiforme 16-Cabo Vírgenes Lepidophyllum –24.9 Aulacomya ater 1.8 10/26/01 1.35 cupressiforme 17-Cañadón Beta Festuca sp. –24.9 Aulacomya ater 0.6 07/15/99 1.50 18-Bahía San Sebastián Berberis sp. –25.0 Mytilus edulis 1.1 07/17/99 1.35 (Playa de Los Chorrillos) 19-Bahía San Sebastián Berberis sp. –25.0 Mytilus edulis 1.1 07/21/99 1.25 (Sitio San Genaro 2) 20-Beagle Channel- Berberis buxifola –25.0 Mytilus edulis 10.6 10/22/01 1.35 Puerto Harberton 21-Isla Livingston Foliose lichens –20.3 Patinigera sp. 2.0 12/28/98 0.90 (Antártida) explain the notorious ∆R increase in this sector. In the neighboring Punta Norte, the contact with the open sea reduces this local effect, providing an age coherent with oceanic parameters. The values for Playa Unión and Punta Loyola indicate a depressed RE despite their contact with the open sea. This may be explained by the influence of continental waters from the Chubut and Gal- legos rivers, respectively, as these localities are found south of their mouths, in the direction the water is dragged by littoral drift. It is suggested that, since they present a better equilibrium with continental waters (dissolved bicarbonate of a fundamentally biogenic origin), the effect is locally diluted. Age Offsets on the Coast of Patagonia (Argentina) 13

Table 2 Preliminary results of the ∆R estimation in the areas considered Lat (S) Locality Long (W) ∆R (14C yr) Lab # Observations 1-Punta Norte 42°12′ 566 ± 80 AC1504 OSa 63°38′ AC1510 2-Golfo San José (Playa 42°20′ 856 ± 105 AC1504 RCb Larralde), surficial water 64°30′ AC1614 PACc 3-Golfo San José (Playa 42°20′ 1120 ± 180 AC1504 RC Larralde), deep water 64°30′ AC1497 PAC 4- Golfo Nuevo 42°23′ 210 ± 35 AC1504 RC (Playa Villarino) 64°30′ AC1585 5- Golfo Nuevo- 42°46′ 260 ± 45 AC1504 RC Puerto Madryn 65°02′ AC1638 6-Playa Unión 43°18′ 230 ± 70 AC1504 OS 65°03′ AC1509 7-Comodoro Rivadavia 45°52′ 415 ± 45 AC1548 OS 67°29′ AC1526 8-Rada Tilly 55°56′ 463 ± 55 AC1607 OS 67°32′ AC1605 9-Caleta Olivia 46°26′ 370 ± 65 AC1548 OS 67°32′ AC1566 10-Isla Lobos 47°56′ 355 ± 50 AC1606 OS 65°51′ AC1639 11-Bahía Desvelos 48°19′ 288 ± 48 AC1606 RC 66°20′ AC1640 12-Bahía Laura 48°24′ 185 ± 35 AC1606 RC 66°28′ AC1586 13-Puerto San Julián 49°19′ 485 ± 60 AC1527 OS 67°42′ AC1528 14-Punta Bustamante 51°34′ 529 ± 90 AC1499 OS 68°58′ AC1498 15-Punta Loyola 51°37′ 90 ± 25 AC1529 CCd 69°01′ AC1530 16-Cabo Vírgenes 52°20′ 516 ± 85 AC1500 OS 68°21′ AC1496 17-Cañadón Beta 52°43′ 555 ± 45 AC1500 OS 68°40′ AC1556 18-Bahía San Sebastián 53°15′ 80 ± 35 AC1502 RC (Playa de Los Chorrillos) 68°28′ AC1511 19-Bahía San Sebastián 53°19′ 265 ± 45 AC1502 RC (Sitio San Genaro 2) 68°17′ AC1565 20-Beagle Channel- 54°53′ 662 ± 65 AC1501 OS Puerto Harberton 67°20′ AC1547 21-Isla Livingston 64°27′ 1900 ± 140 AC1507 OS (Antártica) 61°37′ AC1505 aOS: open sea bRC: restricted connection with the open sea cPAC: presence of aged carbonates dCC: continental water distribution 14 R R Cordero et al.

In coastal entries such as Golfo Nuevo, Desvelos, and the Laura and San Sebastián bays, depressed values would indicate a more effective mixture of seawater with atmospheric CO2. Coastal geometry and the smaller depth of the water in this sector facilitate this fact. Finally, measured values in open-sea localities record a tendency for a latitudinal increase between Comodoro Rivadavia (Chubut) and Puerto Harberton (Tierra del Fuego) according to the global oceanic pattern. The value for Livingston Island (Antarctic) is also coherent with its latitude (Omoto 1983), although we lack contextual information for a better characterization of RE in this area.

SUMMARY AND CONCLUSIONS The correction due to the RE is standardized at 400 yr for the global oceanic average. This correction, however, is not representative of the values corresponding to many coastal areas. The Atlantic Fuego-Patagonic coast, which is more than 3000 km long, presents a significant spectrum of variations in their 14C age offsets. Framed in this variability and waiting for more representative values in each locality, our present research provides information that may indicate the local values of 14C age offsets compared to the global oceanic average. The range of the dispersion of values obtained for these regions, on the other hand, mirrors the limitations in the systematic use of the global sea average for archaeological samples. This emphasizes the importance of local conditions. The determination of 14C age offsets is especially useful for human occupations in the Middle and Late Holocene, where differences in centuries may be important in the interpretation of the cultural chronology. This is even more important for Patagonic archaeology, in which human populations based their economy on the hunter-gatherer lifestyle until the beginning of the 20th century. In this area, archaeological studies have increased in the last years (Gómez Otero 1995), but in the Tierra del Fuego sector research is 3 decades old (Borrero 1994–5; Orquera and Piana 1999), which highlights the increasing importance of improving the chronological tools used. Estimating the value of the local RE is difficult due to uncertainties in calibrating ages for the South- ern Hemisphere, but the approach given by 14C age offsets not only widens the possibilities of cor- relating or comparing sites along the coast, but also allows including in the discussion sites inside the continent, where materials available for dating are not affected by this phenomenon.

ACKNOWLEDGMENTS We want to thank the following sample collectors: Dr Gonzalez Bonorino, Dr Alicia Castro, Dr Patricia Miretzky, Lic Flavia Marina Carballo, Sr Roberto Taylor, and students Ramiro Barberena, Atilio F J Zangrando, Graciela Flores, Vivian Arias, Leticia Raffaele, Lorena Carrera, Tirso Bourlot, Augusto Tessone, Marcia Bianchi Villelli, Javier Musali, and Luciano Paffundi.

REFERENCES Albero M, Angiolini FE, Piana EL. 1986. Discordant Bard E, Arnold M, Fairbanks RG, Hamelin B. 1993. ages related to reservoir effect of associated archaeo- 230Th-234U and 14C ages obtained by mass spectrome- logic remains from the Tunel Site, Beagle Channel, try on corals. Radiocarbon 35(1):191–9. Argentine Republic. Radiocarbon 28(2A):748–53. Borrero LA. 1994–5. Arqueología de la Patagonia. Pal- Albero M, Angiolini FE, Piana EL. 1987. Holocene 14C impsesto. Revista de Arqueología 4:9–69. reservoir effect at Beagle Channel (Tierra del Fuego, Codignotto JO. 1996. Cuaternario y Dinámica Costera. Argentina Republic). Quaternary of South America Geología y Recursos Naturales de la Plataforma Con- and Antartic Peninsula 5:59–71. tinental Argentina. In: Ramos VA, Turic MA, editors. Age Offsets on the Coast of Patagonia (Argentina) 15

XIII Congreso Geológico Argentino y III Congreso de Markgraf V. 1993. Paleoenvironments and paleoclimates Exploración de Hidrocarburos. Relatorio 2:17–28. in Tierra del Fuego and southernmost Patagonia, de Vries H. 1958. Atom bomb effect: variations of radio- South America. Palaeogeography, Palaeoclimatol- carbon in plants, shells, and snails in the past 4 years. ogy, Palaeoecology 102:53–68. Science 128:250–1. Omoto K. 1983. The problem and significance of radio- Dye T. 1994. Apparent ages of marine shells: implica- carbon geochronology in Antarctica. Camberra: Aus- tions for archaeological dating in Hawaii. Radiocar- tralian Academy of Science. p 450–2. bon 36(1):51–7. Orquera LA, Piana EL. 1999. El extremo Austral del Figini AJ. 1999. Comparación de edades C-14 en mues- Continente Nueva historia de la Nación Argentina. tras de origen marino y terrestre. Efecto de reservorio. Academia Nacional de la Historia 1:233–257. Actas del XII Congreso Nacional de Arqueología Ar- Robinson SW, Trimble D. 1981. Natural and man-made gentina (II):353–6. radiocarbon as a tracer for coastal upwelling pro- Gómez Otero J. 1994. Reseña sobre la arqueología en la cesses. In: Richards FA, editor. Coastal Upwelling. Provincia de Chubut. Guía de Campo Península Washington DC: American Geophysical Union. Valdés y Centro Noroeste del Chubut. Séptima re- p 298–302. unión de campo. Puerto Madryn: CADINQUA. p 29– Stuiver M, Braziunas TF. 1993. Modeling atmospheric 43. 14C influences and 14C ages of marine samples to Gómez Otero J. 1995. Bases para una arqueología de la 10,000 BC. Radiocarbon 35(1):137–89. Costa Patagónica Central. Arqueología 5:61–103. Stuiver M, Reimer PJ. 1993. Extended 14C data base and Ingram BL, Southon JR. 1996. Reservoir ages in eastern revised Calib 3.0 14C age calibration program. Radio- Pacific coastal and estuarine waters. Radiocarbon carbon 35:215–30. 38(3):573–82. Stuiver M, Reimer PJ, Braziunas TF. 1998. High-preci- Mancini MV. 1998. Vegetational changes during the Ho- sion radiocarbon age calibration for terrestrial and ma- locene in Extra-Andean Patagonia, Santa Cruz Prov- rine samples. Radiocarbon 40(3):1127–51. ince, Argentina. Palaeogeography, Palaeoclimatol- Suess HE. 1955. Radiocarbon concentration in modern ogy, Palaeoecology 138:207–19. wood. Science 122:415–41. RADIOCARBON, Vol 45, Nr 1, 2003, p 17–24 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

RADIOCARBON DATING OF INDIVIDUAL FATTY ACIDS AS A TOOL FOR REFINING ANTARCTIC MARGIN SEDIMENT CHRONOLOGIES

Naohiko Ohkouchi1,2 • Timothy I Eglinton1,3 • John M Hayes4

ABSTRACT. We have measured the radiocarbon contents of individual, solvent-extractable, short-chain (C14, C16, and C18) fatty acids isolated from Ross Sea surface sediments. The corresponding 14C ages are equivalent to that of the post-bomb dissolved inorganic carbon (DIC) reservoir. Moreover, molecular 14C variations in surficial (upper 15 cm) sediments indicate that these compounds may prove useful for reconstructing chronologies of Antarctic margin sediments containing uncertain (and potentially variable) quantities of relict organic carbon. A preliminary molecular 14C chronology suggests that the accumulation rate of relict organic matter has not changed during the last 500 14C yr. The focus of this study is to determine the validity of compound-specific 14C analysis as a technique for reconstructing chronologies of Antarctic margin sediments.

INTRODUCTION The stability of the West Antarctic Ice Sheet (WAIS) is thought to be sensitive to global warming because a major part of this ice sheet is grounded below sea level (Mercer 1978; Alley and Whillans 1991; Oppenheimer 1998). To investigate its behavior in relation to climatic change, many studies have focused on past variability of the WAIS over the past glacial-interglacial cycle. Since Antarctic margin sediments generally lack calcareous foraminifera, acid-insoluble organic carbon (AIOC) has often been used to establish radiocarbon chronologies for sediments deposited over the late Quater- nary (Domack et al. 1989; Andrew et al. 1999; Licht et al. 1996; Harris et al. 1996). However, Ant- arctic margin sediments commonly contain variable amounts of reworked sediment eroded from the Antarctic continent (Sackett et al. 1974). This “contamination” by relict organic carbon (OC) leads to anomalously old core-top ages or to age reversals down-core, which leaves the reconstruction of oceanic environments around Antarctica far behind that of the Laurentide Ice Sheet (e.g., Bond and Lotti 1995). In an attempt to overcome this problem, we applied compound-specific 14C dating (Eglinton et al. 1996, 1997) to a suite of Ross Sea sediments. Pearson et al. (2000) reported that individual sterols in laminated sediments from the Santa Monica Basin served as effective tracers of surface water dissolved inorganic carbon. A goal of this paper is to determine the 14C contents of different fatty acids with a view toward identifying those suitable for reconstruction of sediment chronologies.

METHODS In the Ross Sea, sediments were collected using a box corer from the Chinstrap (76°19.9′S, 165°01.5′E, water depth 827 m), Gentoo (76°20.4′S, 172°56.2′E, 623 m), Emperor (76°58.9′S, 171°59.7′E, 670 m), and Fairy (77°58.3′S, 178°03.0′W, 671 m) sites during the ROAVERRS (Research on Ocean-Atmosphere Variability and Ecosystem Response in the Ross Sea program) cruise on the R/V Nathaniel B Palmer in December 1998. The Chinstrap site represents a region where dia- toms bloom during the austral summer, whereas the Emperor and Gentoo sites are located in a region of extensive blooms of haptophyte algae. The Fairy site is a location where large algal blooms have not been observed. Therefore, the sediments analyzed in this study span a range of surface ocean conditions in the Ross Sea. These sediments were stored in a freezer (–20 °C) until analysis.

1Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, USA. 2Corresponding author. Email: [email protected]. 3Email: [email protected]. 4Department of Marine Geology and Physics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, USA. Email: [email protected].

17 18 N Ohkouchi et al.

Figure 1 Map showing study area and sample locations. The depth contour was derived from The Generic Mapping Tools (GMT; www.gmt.soest.hawaii.edu) and corresponds to 50-m intervals.

Dried sediments (100–350 g) were Soxhlet-extracted for 4 days with dichloromethane/MeOH (93: 7, v/v). The extracts were saponified by 0.7M KOH/MeOH and esterified by HCl/MeOH under a N2 atmosphere. They were separated into fractions by pressurized flash chromatography. Compounds were quantified using a HP5890 gas chromatograph (GC) equipped with CPSil-5CB column (60 m × 0.25 mm i.d., film thickness 0.25 µm) and FID. Individual compounds were isolated and purified using methods described elsewhere (Eglinton et al. 1996). Briefly, samples containing fatty acid methyl esters (FAMEs) were repeatedly injected to a preparative-capillary GC equipped with Ger- stel CIS injector and CPSil 5CB column (100 m × 0.52 mm i.d., film thickness 0.25 µm) and individual FAMEs were trapped from the column effluent. Microcolumn SiO2 chromatography (hexane/ dichloromethane, 2:1, v/v) was used to remove column bleed and the individual compounds were transferred to quartz tubes for combustion. The resulting CO2 was converted to graphite (Pearson et al. 1998). The purity of each compound was determined based on the GC-FID analysis of an aliquot of the isolated component. 14C content in the samples was determined at the National Ocean Science Accelerator Mass Spectrometry (NOSAMS) Facility at the Woods Hole Oceanographic Institution. 14 14 All C values are corrected for the contribution of methyl carbon obtained from MeOH ( CMeOH = –955.6 ‰) during the esterification by isotope mass balance. Dating Fatty Acids to Refine Antartic Chronologies 19

RESULTS AND DISCUSSION Concentrations of solvent-extractable fatty acids (FAs) in the Ross Sea surface sediments show a bimodal pattern with peaks at C16 and C24 or C26 and have a strong predominance of even carbon- number chains (Figure 2). Although sediments from the Chinstrap site are enriched in FAs relative to other sites, relative abundances of FAs to the total organic carbon (TOC) content are rather similar between sites. Concentrations of short-chain (C14, C16, and C18) FAs decrease with depth in the Chinstrap site. The total concentrations of short-chain FAs in shallower samples (0–2 and 3–6 cm) are 8.80 and 8.78 µg/g dry sediment (ds), respectively, whereas in deeper samples (6–9, 9–12, and 12–15 cm) they are lower than 3.5 µg/g ds. In contrast, total concentrations of long-chain (C24, C26, and C28) FAs for all samples fall in a narrow range from 2.0 to 3.2 µg/g ds. Since TOC contents of these sediments vary little with depth, the reduction of the short-chain FAs may reflect their lability compared with the TOC.

Figure 2 Concentrations of solvent-extractable C12–C32 fatty acids relative to sediment dry weight in surface sediments from the Chinstrap, Gentoo, Emperor, and Fairy sites. Total organic carbon contents of these sediments are also shown.

Figure 3 Concentrations of solvent-extractable C12–C32 fatty acids relative to sediment dry weight in down-core sediments from the Chinstrap site. Total organic carbon contents of these sediments are also shown. 20 N Ohkouchi et al.

As shown in Figure 4, the short-chain FAs are significantly enriched in 14C relative to the AIOC. With the exception of the Fairy site, the ∆14C values of the short-chain FAs are relatively uniform and mostly fall in the range of the post-bomb dissolved inorganic carbon (DIC) reservoir (–100 ± 20‰) in this region (Berkman and Forman 1996; Gordon and Harkness 1992). Since these short-chain FAs are produced by not only marine algae but also by other organisms (including animals and bacteria) as membrane components, energy storage material, heat insulating material, etc. (Harwood 1996), some fraction of these FAs could have been produced by these heterotrophic organisms. Whatever their immediate sources are, our results indicate that the carbon in the short-chain FAs is derived ulti- mately from the photosynthetic fixation of surface ocean DIC within the last 40 yr, during which levels of atmospheric 14C have been elevated by the testing of weapons. An important implication is that 14 contributions of C14, C16, and C18 FAs from relict, C-free OC eroded from the Antarctic continental erosion are not significant at these sites. If this dominance of autochthonous sources prevails down- core, 14C analyses of solvent-extractable, short-chain FAs can provide a basis for accurate sediment chronologies. At the Fairy site, 14C values of the short-chain FAs are 70–150‰ lower than that of the post-bomb DIC. Although we do not have confirming evidence, deep bioturbation and/or a low sedimentation rate may cause the reduced 14C level in the FAs.

Figure 4 Radiocarbon activities (as 14C) of individual fatty acids (circles) and acid-insoluble organic carbon (AIOC; squares) in the surface (0–2 cm) sediments collected from the Chinstrap, Emperor, Gentoo, and Fairy sites in the Ross Sea. Numbers in the diagram indicate carbon numbers of fatty acids. The error bars represent ±1 standard deviation. Shaded areas indicate 14C activities of DIC reservoir in the Ross Sea during the post-bomb (after 1957) and pre-bomb (before 1957) eras (Berkman and Forman 1996; Gordon and Harkness 1992). Dating Fatty Acids to Refine Antartic Chronologies 21

In marked contrast to the short-chain FAs, the long-chain FAs are substantially more depleted in 14C but generally enriched relative to the AIOC. In general, the shorter the chain length, the younger the age. Even carbon-numbered chains are strongly predominant among the long-chain FAs at these sites, suggesting derivation from unaltered biological debris. The distribution is not unlike that in waxes from the leaves of higher plants (Eglinton and Hamilton 1967) but eolian transport would then be indicated and a uniform distribution among sites would be expected (Ohkouchi et al. 2000). A relationship to the long-chain FAs found in soils in Antarctic dry valleys (Matsumoto et al. 1981), which could also be delivered by wind erosion, is unlikely for heterogeneous FA patterns because the even- C/odd-C concentration ratios in those soil FAs are much lower than observed in the sediments. The observed variations in ∆14C (Figure 4) indicate (i) the presence of at least some recently- produced, long-chain FAs, presumably of algal or zooplanktonic origin (Volkman et al. 1998), and (ii) significant variations in the modern/relict mixing ratio for these compounds. To check the usefulness of the 14C content of the short-chain FAs as chronological tools at greater 14 depths of the sediments, a vertical profile of C content of C16 FA was examined for 5 depth intervals from the upper 15 cm of a box core from the Chinstrap site (Figure 5). Essentially constant 14C

14 Figure 5 Depth-related variations in C ages of the C16 fatty acid (circles) and acid-insoluble organic carbon (AIOC; squares) in the down-core sediments from the Chinstrap site. In the depth 14 range 6–9 cm, a mixture of C14 and C16 fatty acids was measured. The error bars of the C age represent a ±1 standard deviation. Shaded areas indicate the post-bomb and pre-bomb DIC reservoir ages (Berkman and Forman 1996; Gordon and Harkness 1992). 22 N Ohkouchi et al. ages are evident for the samples from the upper 6 cm, corresponding to the depth of the bioturbated mixed layer. The ages of these 2 samples are nearly identical to the post-bomb DIC reservoir age. The age of the C16 FA in the mixed layer mainly reflects the balance between the bioturbation rate and degradation rate of this compound. Using a model analogous to the CaCO3 dissolution and mixing (Broecker et al. 1991), the 14C age of the FA will correspond with the post-bomb DIC reservoir age if the FA reaching the sea floor is rapidly (within a few yr) mixed and if its half-life in the mixed layer is less than 30 yr. The latter is a reasonable assumption based on studies of FA degradation in marine sediments (Canuel and Martens 1996). In Antarctic margin sediments, such as those underlying the Ross Sea, elevated bioturbation rates may be caused by the high activity of benthos, supported by the extremely high flux of organic detritus (DiTullio et al. 2000). The 14C ages of AIOC 14 roughly parallel that of the C16 FA with an offset of 1200–2000 C yr. Sedimentation rates estimated from C16 FA and AIOC chronologies are around 7.5 and 15 cm kyr–1, respectively. The two-fold difference between FA- and AIOC-based rates of sedimentation could be explained if the fraction of relict AIOC increases with depth. Alternatively, it could indicate that with increasing depth progres- sively greater portions of the C16 FA were derived from relict carbon, presumably by bacterial attack on imported kerogen (Petsch et al. 2001). Assuming AIOC the in Ross Sea sediments is mainly derived from 2 sources, marine autochthonous and relict, we can estimate the relative contribution of relict OC from the isotopic difference between the short-chain FAs and AIOC. Terrestrial OC transported from continents other than Ant- arctica through the atmosphere should be a minor component. Assigning the ∆14C value of autoch- 14 14 thonous organic matter as that of the C16 FA and ∆ C of relict organic matter of infinite C age (∆14C = –1000‰), we calculated the minimum contribution of relict OC to the AIOC (Figure 6). At the Emperor and Fairy sites, the contributions of relict OC are higher than 50%. At the Chinstrap site, the contribution of relict OC is relatively invariant, comprising about 20% of the total AIOC. The relative abundance of relict OC could be a function of variable input of debris-laden glacial ice in the overlying water column. Our record suggests that the input of relict sediments to the Ross Sea (potentially related to the melting rate of glacial ice) did not change substantially during the last 600 14C yr.

Figure 6 Estimated minimum fraction of relict organic carbon (shaded area) in the acid-insoluble organic carbon (AIOC) in the surface sediments from the Emperor, Gentoo, Fairy, and Chinstrap sites, based on the two-source model explained in the text. Dating Fatty Acids to Refine Antartic Chronologies 23

Overall, this initial application of compound-specific 14C dating to Antarctic margin sediments clearly shows a potential utility of this approach for developing sediment chronologies. In addition to FAs, sterols and isoprenoid alkenes, which are also derived from phytoplankton (Nichols et al. 1988), have been identified in Antarctic margin sediments and could prove as useful for 14C dating (Pearson et al. 2000). The short-chain FAs are particularly useful because they are among the most abundant lipid-class compounds in the Antarctic margin sediments (e.g., Venkatesan and Kaplan 1987). Since prior studies of sediment cores have yielded core-top 14C ages of AIOC of 2000–4000 yr BP (Licht et al. 1998), the timing of WAIS recession and/or advance estimated from these dates may be overestimated by 1000–3000 14C yr. Substantially improved chronologies can be obtained if the age scale is corrected for the core-top age but this approach may be prone to error because it assumes a constant age offset downcore. Without the application of appropriate corrections for relict OC, molecular 14C measurements provide an effective way of eliminating such interferences.

ACKNOWLEDGMENTS We thank J Grebmeier and J P Barry for providing samples, and J Andrews and K Licht for encouraging us to undertake this study. We also thank A P McNichol and the NOSAMS staff for 14C measurements, A Pearson for technical advice, and D Montlucon for assistance in the laboratory. This research was supported by the grants from NSF (OCE-9907129, OCE-9809624, and OPP- 9909782), and Japan Society for the Promotion of Science.

REFERENCES Alley RB, Whillans IM. 1991. Changes in the West Ant- Domack EW, Jull AJT, Anderson JB, Linick TW, Will- arctic Ice Sheet. Science 254:959–63. iams CR. 1989. Application of tandem accelerator Andrews JT. 1999. Problems and possible solutions con- mass-spectrometer dating to late Pleistocene-Ho- cerning radiocarbon dating of surface marine sedi- locene sediments of the east Antarctic continental ments, Ross Sea, Antarctica. Quaternary Research 52: shelf. Quaternary Research 31:277–87. 206–16. Eglinton G, Hamilton RJ. 1967. Leaf epicuticular waxes. Berkman PA, Forman SL. 1996. Pre-bomb radiocarbon Science 156:1322–35. and the reservoir correction for calcareous marine spe- Eglinton TI, Aluwihare LI, Bauer JE, Druffel ERM, Mc- cies in the Southern Ocean. Geophysical Research Nichol AP. 1996. Gas chromatographic isolation of in- Letters 23:363–66. dividual compounds from complex matrices for radio- Bond G, Lotti R. 1995. Iceberg discharges into the North carbon dating. Analytical Chemistry 68:904–12. Atlantic on millennial timescales during the last de- Eglinton TI, Benitez-Nelson BC, Pearson A, McNichol glaciation. Science 267:1005–266. AP, Bauer JE, Druffel ERM. 1997. Variability in ra- Broecker WS, Klas M, Clark E, Bonani G, Ivy S, Wolfli diocarbon ages of individual organic compounds from W. 1991. The influence of CaCO3 dissolution on core- marine sediments. Science 277:796–9. top radiocarbon ages for deep-sea sediments. Pale- Gordon JE, Harkness DD. 1992. Magnitude and geo- oceanography 6:593–608. graphic variation of the radiocarbon content in Antarc- Canuel EA, Martens CS. 1996. Reactivity of recently de- tic marine life: implications for reservoir corrections posited organic matter: degradation of lipid com- in radiocarbon dating. Quaternary Science Reviews pounds near the sediment-water interface. Geochim- 11:697–708. ica et Cosmochimica Acta 60:1793–806. Harris PT, O’Brien PE, Sedwick P, Truswell EM. 1996. DeMaster DJ, Ragueneau O, Nittouer CA. 1996. Preser- Late Quaternary history of sedimentation on the Mac. vation efficiencies and accumulation rates for bio- Robertson Shelf, east Antarctica: problems with 14C genic silica and organic C, N, and P in high-latitude dating of marine sediment cores. Papers and Proceed- sediments: the Ross Sea. Journal of Geophysical Re- ings of the Royal Society of Tasmania 130:47–53. search 101:18501–18. Harwood JL. 1996. Recent advances in the biosynthesis DiTullio GR, Grebmeier JM, Arrigo KR, Lizotte MP, of plant fatty acids. Biochemica et Biophysica Acta Robinson DH, Leventer A, Barry JP, van Woert ML, 1301:7–56. Dunbar RB. 2000. Rapid and early export of Phaeo- Licht KL, Jennings AE, Andrews JT, Williams KM. cystis Antarctica blooms in the Ross Sea, Antarctica. 1996. Chronology of late Wisconsin ice retreat from Nature 404:595–8. the western Ross Sea, Antarctica. Geology 24:223–6. 24 N Ohkouchi et al.

Licht KJ, Cunningham WL, Andrews JT, Domack EW, tracer for surface ocean radiocarbon. Paleoceanogra- Jennings AE. 1998. Establishing chronologies from phy 15:541–50. acid-insoluble organic 14C dates on Antarctic (Ross Petsch ST, Eglinton TI, Edwards KJ. 2001. 14C dead liv- Sea) and Arctic (North Atlantic) marine sediments. ing biomass: evidence for microbial assimilation of Polar Research 17:203–16. ancient organic carbon during shale weathering. Sci- Matsumoto G, Torii T, Hanya T. 1981. High abundances ence 292:1127–31. of long-chain normal alkanoic acids in Antarctic soil. Pearson A, McNichol AP, Schneider RJ, von Reden KF. Nature 290:688–90. 1998. Microscale AMS 14C measurement at Mercer JH. 1978. West Antarctic ice sheet and CO2 NOSAMS. Radiocarbon 40:61–75. greenhouse effect: a threat of disaster. Nature 271: Pearson A, Eglinton TI, McNichol AP. 2000. An organic 321–5. tracer for surface ocean radiocarbon. Paleoceanogra- Nichols PD, Volkman JK, Palmisano AC, Smith GA, phy 15:541–50. White DC. 1988. Occurrence of an isoprenoid C25 di- Sackett WM, Poag CW, Eadie BJ. 1974. Kerogen recy- unsaturated alkene and high neutral lipid content in cling in the Ross Sea, Antarctica. Science 185:1045–7. Antarctic sea-ice diatom communities. Journal of Stuiver M, Polach HA. 1977. Discussion: reporting of Phycology 24:90–6. 14C data. Radiocarbon 28:355–63. Ohkouchi N, Kawamura K, Takemoto N, Ikehara M, Na- Venkatesan MI, Kaplan IR. 1987. The lipid geochemistry katsuka T. 2000. Implications of carbon isotope ratios of Antarctic marine sediments: Bransfield Strait. Ma- of C27–C33 alkanes and C37 alkenes for the sources of rine Chemistry 21:347–75. organic matter in the Southern Ocean surface sedi- Volkman JK, Barrett SM, Blackburn SI, Mansour MP, ments. Geophysical Research Letters 27:233–6. Sikes EL, Gelin F. 1998. Microalgal biomarkers: a re- Oppenheimer M. 1998. Global warming and the stability view of recent research developments. Organic of the West Antarctic Ice Sheet. Nature 393:325–32. Geochemistry 29:1163–79. Pearson A, Eglinton TI, McNichol AP. 2000. An organic RADIOCARBON, Vol 45, Nr 1, 2003, p 25–32 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

14C CHRONOLOGY OF LATE PLEISTOCENE–HOLOCENE EVENTS IN THE NIZHNEE PRIAMURIE (SOUTHEAST RUSSIA)

V B Bazarova1 • L M Mokhova1 • L A Orlova2 • M A Klimin3 • I G Gvozdeva4

ABSTRACT. The Russian Far East is characterized by widespread peat bogs with a sufficiently thick peat accumulation. A series of radiocarbon dates from the studied peat bogs (in Lower Amur) were obtained. Analysis of these dates shows that the total peat formation in this territory began in the Late Pleistocene–Holocene (11830 ± 820, TIG-157; 9975 ± 120, SOAN- 4025). The rates of peat accumulation and the humidity index were counted. In addition, the botanical composition and degree of peat decomposition were defined. These data allow to study in more detail climate fluctuation and the 14C chronology of Holocene events in the region studied.

INTRODUCTION Due to continuous accumulation, a peat sequence is a good source for the study of peat bog formation conditions, dynamics of swamping, and the details of climate fluctuation. The continuous accumulation of peat allows us to reconstruct an amplitude of climatic changes and provides a better description of the paleogeographic boundaries in the Late Pleistocene and Holocene. According to the Peat Fund USSR’s division of peat land into districts, the Nizhnee Priamurie territory was distinguished as the largest peat area in the Russian Far East. Peat lands and bogs take up about 58,000 km2 (Prozorov 1970) and spread to the Amuro-Amgunskaya, Udil-Kizinskaya, Evoron-Chugchagirskaya, and Credneamurskaya lake-alluvial depressions. Using the data of Neishtadt (1967) and Sokhina et al. (1978), the development of 4 vegetal phases were determined in the Holocene in the territory of the Sredniy and Nizhnee Priamuries. The first phase (Earliest Holocene) consisted of small-leaved forests and shrubs. The second phase (Early Holocene) consisted of coniferous forests, along with small- and broad-leaved trees. The third phase (the Quercus mongolica phase—Middle Holocene) was defined by coniferous, broad- and small- leaved forests. The fourth phase (Pinus koraiensis phase—Late Holocene) was characterized by coniferous, small-leaved forests and broad-leaved trees. Pollen data from the Gurskiy peat bog are described by Korotky et al. (2000). The climate of the studied area is temperate with monsoon traits. During the year, the meridional circulation (caused by southern tropical cyclones penetrating far into the north) predominates in this territory and brings a dry, cold air. The average annual temperature in this territory is negative °C. The coldest month is January (–27 to –22 °C) and the warmest month is July (16–18 °C). The precipitation varies drastically during the year. In the warm months, the precipitation in the region is about 80–90% of its total yearly rainfall, while in the winter the rainfall is only about 10–20% of the yearly total. The annual sum of precipitation fluctuates from 500–550 mm in the Sredneamurskaya depression up to 1000–1100 mm on the Tatar Strait coast. The large quantity of precipitation, occur- ing in a short period, is one of the main reasons for swamping in this territory. The humidity coeffi- cient ranges from 1.4 up to 2.8, characterizing this territory as sufficiently moist and favorable for widespread bogs.

1Pacific Institute of Geography, Radio St. 7, Vladivostok 690041, Russia. 2United Institute of Geology, Geophysics and Mineralogy, Universitetsky Pr. 3, Novosibirsk 630090, Russia. 3Institute of Water and Ecological Problems, Kim-Yu-Chen St. 65, Khabarovsk 680063, Russia. 4Pacific Institute of Oceanology, Balti’skaya St. 43, Vladivostok 690041, Russia.

25 26 V B Bazarova et al.

Up to now, only a few 14C dates were obtained for peat bogs in southeast Russia (Neishtadt 1957; Kulakov et al. 1975; Khotinskiy 1977; Sokhina et al. 1978; Prozorov 1985; Mikishin et al. 1987). Such a small number of dates do not allow the authors to correlate in detail the Holocene paleogeographic events in this territory. Two peat bogs in the Nizhnee Priamurie were studied. The Gurskiy peat bog is situated in the northwestern region of the Sredneamurskaya depression. The absolute altitude of this peat bog is 35 m; the thickness of the deposits is 3.5 m. The Tyapka peat bog is situated in the northern region of the Amuro-Amgunskaya depression, near the Tyapka river. The absolute altitude of this peat bog is 40 m; the thickness of the peat bog is 5.4 m (Figure 1).

Figure 1 Map showing the location of study sections: 1=Gurskiy; 2=Tyapka

METHODS Paleoenvironmental reconstructions were based on detailed 14C dating, pollen, and botanical data. Also, we defined the degree of peat decomposition, counted rates of peat accumulation, and determined the index of humidity for the 2 regions. For 14C dating, the peats were first cleaned by acid/ alkali/acid. The samples were then decalcified using a hot 3% HCl solution and humics were extracted using a 1 N KOH solution. The extracted matter was acidified to pH 1.0 with dilute HCl to recover the humic fraction. The 14C activity was measured by liquid scintillation counting and the index of humidity was derived by the Elina method for both regions (Elina et al. 1992). Results of all analyses are presented in Tables 1–3. 14C Chronology of Nizhnee Priamurie (Southeast Russia) 27

Table 1 14C dates for Gurskiy and Tyapka sections Gurskiy section Tyapka section Depth (m) 14C age Depth (m) 14C age 0.5 2720 ± 120 0.50 1395 ± 60 (TIG-151) (SOAN-4017) 1.0 4600 ± 200 1.00 2200 ± 70 (TIG-152) (SOAN-4018) 1.5 6880 ± 270 1.30 2025 ± 100 (TIG-153) (SOAN-4162) 2.1 8540 ± 320 1.70 2480 ± 70 (TIG-154) (SOAN-4163) 2.6 9180 ± 350 1.90 2730 ± 65 (TIG-155) (SOAN-4019) 3.1 10220 ± 750 2.55 3510 ± 60 (TIG-156) (SOAN-4044) 3.5 11830 ± 820 2.75 4270 ± 125 (TIG-157) (SOAN-4021) 3.05 5325 ± 95 (SOAN-4022) 3.55 6240 ± 100 (SOAN-4023) 4.20 7720 ± 100 (SOAN-4024) 5.20 9975 ± 120 (SOAN-4025) 5.30 9910 ± 50 (AA-36302)

RESULTS AND DISCUSSION The age of the lowest layer of the Gurskiy peat bog is 11830 ± 820 BP (TIG-157) (Table 1). It is one of the oldest among the peat bogs in southeast Russia. In this region, the peat bog near Chlya Lake has the oldest age (Mikishin et al. 1987). The peat accumulation in this peat bog began in the Late Pleistocene and was continuous during the Holocene. The lowest part of the Gurskiy peat deposits were formed at the base of a hill, where thickness of the peat varies from 3.1 up to 3.7 m, whereas the thickness of peat deposits around this hill base is about 2.7 m. Thus, the local geomorphological features of this place and microclimate were the main causes for the primary peat accumulation in the Late Pleistocene. After that, swamping began in the surrounding territories at the beginning of the Early Holocene. The history of the Gurskiy peat bog development illustrates the role of both climatic zone conditions and local climatic conditions, which sometimes are more considerable. The 14C date of 10220 ± 750 BP (3.1 m depth) corresponds to the boundary of the Pleistocene– Holocene, and the first appearance of the post-glacial forests in Europe, Japan, and the Pacific coast of Alaska (Khotinskiy 1977). The 14C date of 4600 ± 200 BP (1 m depth) characterizes the boundary of the Atlantic and Subboreal phases of the Holocene and the 14C date of 2720 ± 120 BP (0.5 m depth) is near the boundary of the Subboreal and Subatlantic phases. The peat layer from the 1.5 m depth corresponds to the Atlantic phase and the 2 lower layers situated at depths of 2.1 and 2.6 m, respectively, correspond to the Boreal phase. 28 V B Bazarova et al.

The 14C dates from the Tyapka region show that the peat accumulation in this peat bog began in the Early Holocene, which is later than in the Gurskiy peat bog (Table 1). The 14C date of 7720 ± 100 BP (4.2 m depth) is near the boundary of the Boreal and Atlantic phases. The boundary of the Atlan- tic and Subboreal phases is at 2.75 m depth (4270 ± 125 BP), while the date of 2480 ± 70 BP (1.7 m depth) is the boundary of the Subboreal and Subatlantic periods. The peat layers at the 3.05 and 3.55 m depths correspond to the Atlantic period and the layer at 5.30 m depth corresponds to the Prebo- real period of the Holocene (9975 ± 120 BP). The obtained 14C dates allowed dividing the Holocene into the phases. The sort, accumulation average rate, degree of decomposition, and botanical composition of peat depend on climatic conditions. These data are presented in Tables 2–3. On the basis of these data, we reconstructed the climatic changes in all phases of the Holocene. The description of climatic change is presented in the last columns of Tables 2–3. Figure 2 shows the climatic curves for the study sections and chronological model curve for the northern Eurasia tundra and forest zones (Khotinskiy 1987).

Figure 2 Climatic curves: 1=chronological model curve for northern Eurasia tundra and forest zones (Khotinskiy 1987); 2=for Gurskiy section; 3=for Tyapka section

The general features of these curves are the same. Climate in the Nizhnee Priamurie region changed less abruptly than in the northern part of Eurasia. In Khotinsky’s opinion (1977), the Holocene optimum in the far eastern part of Russia becomes apparent in the Boreal phase. Our data do not confirm this opinion. The climate became warm in the Boreal phase and was warmer than at the present time. 14C Chronology of Nizhnee Priamurie (Southeast Russia) 29 dampest, less cold then in DR-3 damp, warm temperate, damp and warm dampest, temperate warm temperate damp, warm warm dry, temperate damp, cooler then in SBO I temperate damp, III cooler then in AT temperate damp, less II warm, then in AT temperate damp, warmest warm dry, damp, warmer then in BO I dampest, cool damp, cool damp, very cold very damp, Betula Ulmus and , a few quan- , decrease of Quercus nanae Betula nanae, Alnaster Betula, Alnaster tities of Pinus, Betula, Quercus Pinus, Betula, Alnus, Quercus Betula, Alnus, Quercus Betula, Alnus, Quercus Betula, Alnus Alnus Betula, Quercus, Ulmus Betula, Quercus, Ulmus, Betula Quercus, Ulmus Betula, Quercus, Betula, Alnus, Ulmus Betula, Alnus Betula, Alnus Alnaster Index of humidity Vegetation Climate up to 80 3 60 3 Average degree Average of peat decomposition, % 7075 2.5 60 2.5 556070 3 3.5 3 3.5 grassy- sphagnous sphagnous grassy grassy-mossy grassy-mossy real-sphagnous real-sphagnous real-sphagnous moss- sphagnous Average rate Average of peat accumulation sort Peat DR-3 0.25 green moss- AL — moss- green Holocene phases SAT IIISAT — II SAT 0.20 ISAT SBO III — moss green SBO II — arboreal-grassy 0.26SBO I up to 60 20 IIIAT arboreal-grassy — 55 arboreal-grassy II arboreal-grassy AT — 60 3.5 55 IAT 0.29 2.5 BO III arboreal- BO II — arboreal- 0.78 2.5 BO I — arboreal-grassy 2 2.5 up to 85PBO II — arboreal-grassy arboreal-grassy 0.48PBO I 80 75 grassy-arbo- 2.5 — grassy-arbo- grassy-arbo- 2 grassy-green 2 — warm dry, Table 2 Table Environmental characteristics of the Gurskiy section 30 V B Bazarova et al. dry, cool dry, temperate damp, very cool temperate damp, cool dampest, very cool damp, cool temperate dry, cold damper and less I warm than in AT damp, warmest temperate damp, cool dry, very cool dry, cold damper, ; , (up to Alnaster , Betula Betula Alnaster Selaginella Alnaster , Alnaster Alnaster Alnaster Betula nanae Alnaster and and and , decreasing , , decreasing broad-leaved Alnaster Betula and , increasing broad-leaved warm dry, , the first appearance of broad- (up to 40%), apperance of , appearance of broad-leaved (up to 30%), decreasing (up to 37%), maximum ecreasing coniferous, increasing nanae Betula nanae Betula nanae Betula nanae Betula Picea Betula nanae Picea disappearance of broad-leaved Betula 18%), Picea quantity of broad-leaved Betula leaved Betula nanae sharp decrease of Index of humidity Vegetation Climate 10 5 decreasing coniferous, increasing 25up to 30 5 3.5 increasing coniferous, decreasing Average degree of Average peat decomposition, % 20 3 up to 50up to 40 3 60 6.5up to 60 coniferous (up to 19%), up to 60 5 4 3 small increase of coniferous, arboreal sphagnous nous-grassy-arboreal sphagnous arboreal sphagnous sphagnous sphagnous Average rate Average of peat accumulation Peat sort SAT IISAT 0.62 ISAT 0.88 sphagnous-grassySBO III up to 0.80 sphagnous-grassySBO II 20 0.77 sphagnous-grassySBO I 5 0.39 sphagnous-grassy- arboreal-grassy- 4 7 increasing coniferous, decreasing d Table 3 Table region Environmental characteristics of the Tyapka Holocene phases IIISAT 0.36 green moss-sphag- AT III AT 0.20 II AT 0.66 arboreal-grassy- I AT 0.41 sphagnous-grassy- BO II 0.31 arboreal-grassy- BO I 0.31 arboreal-grassy- arboreal-grassy- PBO — sphagnous-grassy up to 40 6 14C Chronology of Nizhnee Priamurie (Southeast Russia) 31

However, the temperature was increasing and had reached the maximum in the Atlantic phase. In the Tyapka section, the climatic optimum of the Holocene is recorded in the first part of the Atlantic phase. In the Gurskiy section, the climatic optimum is recorded in the second part of the Atlantic phase. Thus, the optimum of the Holocene on the coastal region began earlier than in the continental region of the Nizhnee Priamurie. Most investigators observed that there was only 1 cooling in the Holocene in the Nizhnee Priamurie. According to Karaulova (1974), there were 3 coolings in the Holocene in the Russian Far East. Our data indicate 2 coolings on a climatic curve for the Gurskiy section (Figure 2). One of them is in the PBO-BO I phase and the next is in the SAT I phase. There are 3 coolings on the curve for the Tyapka region. The first of them is in the PBO-BO phase and the next 2 coolings are at the beginning of the phase and divide these Subatlantic coolings (the last parts of the Subatlantic phase). There was weak warming in the SBO II phase in the Gurskiy region. However, later in the phase, the glacial cooling (Dryas III) was severe. In addition, the Alleroed warming was warmer in the Gurskiy region than in the continental part of Eurasia (Khotinskiy 1987).

CONCLUSION The detailed 14C dating of peat bogs allowed division of the Holocene into phases for the Nizhnee Priamurie region. The use of particular methods for peat analysis allowed the authors to reconstruct climatic changes in more detail. Local geomorphological features and microclimate were the main causes for the primary peat accumulation about 12000–13000 yr ago. This leads to the conclusion that the Alleroed warming was warmer than on the continental part of Eurasia and the Dryas III cooling was very cold. The active, widespread peat formation in the Nizhnee-Amur region began in the Early Holocene (9000–10000 BP). Climatic and vegetation changes were more slight than on the continent. Holocene climatic changes had a metachronical character in this region. The Boreal phase was warmer than the present climate. On the marine coast, the Holocene optimum is recorded at the beginning of the Atlantic phase, while in the continental part of this region the optimum is recorded in the middle of the Atlantic phase. There were 2 coolings in the continental region and 3 coolings in the coastal region during the Holocene in the Nizhnee Priamurie.

ACKNOWLEDGMENTS This study was supported by the Russian Fund of Fundamental Investigation, grant 00-05-64874.

REFERENCES Elina GA, Yurkovskaya TK.1992. Methods of palaeohy- Korotkiy AM, Mokhova LM, Bazarova VB. Forthcom- drological regime determination as basis for objectiv- ing. Stratigraphic and paleogeographic interpretation ization of causes of mire vegetation successions. Bo- of Gurskiy peat-land. Bulletin of committee by study of tanical Journal 77(7):120–4. In Russian. Quaternary 64. Karaulova LP. 1974. Pollen basis of Pleistocene-Ho- Kulakov AP, Nazarenko EM, Pushkar VS. 1975. About locene deposits stratigraphy in Primorye. traces of Quaternary marine transgressions in North- Khotinskiy NA. 1977. Holocene of Northern Eurasia. ern-Western Priokhotie. Problems of Geomorphology Moscow: Nauka. 197 p. In Russian. and Quaternary Geology in the south of Far East. Khotinskiy NA 1987. Radiocarbon chronology and cor- Vladivostok: FESC AS USSR:118–30. In Russian. relation of natural and anthropogenic boundaries of Mikishin Yu A, Petrenko TI, Gvozdeva IG, Razova GG. Holocene. New data of Quaternary geochronology. 1987. Stratigraphy of deposits of a five meter terrace Moscow: Nauka. p 39–45. In Russian. on Chlya Lake. Pollen data of the Russian East. Vladi- Korotkiy AM, Volkov VG, Bazarova VB, Kovalukh NN. vostok: FESC AS USSR:94–101. In Russian. 1993. Catalogue of Quaternary radiocarbon dates of Neishtadt MI. 1957. History of forests and USSR paleo- Far East. Pacific Institute of Geography. Vladivostok: geography in Holocene. Moscow: Nauka. 404 p. In FEB RUS. 40 p. In Russian. Russian. 32 V B Bazarova et al.

Neishtadt MI. 1967. Stratigraphy of peatbogs by data of Prozorov Yu S. 1985. Peat bogs of the Nizhnee Amur de- absolute ages. The nature of the peat bogs and the pressions. Novosibirsk: Nauka. 212 p. In Russian. methods of the investigations. Leningrad: Nauka. In Sokhina AN, Boyarskaya TD, Okladnikov AP, Roslik- Russian. ova VI, Chernyuk AI. 1978. Section of Quaternary de- Prozorov Yu S. 1970. Total characteristic of peat bogs. posits of the Nizhnee Priamurie. Moscow: Nauka. 107 The resources of surface waters. The Far East 18: p. In Russian. 454–62. In Russian. RADIOCARBON, Vol 45, Nr 1, 2003, p 33–39 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

ON THE COEXISTENCE OF MAN AND EXTINCT PLEISTOCENE MEGAFAUNA AT GRUTA DEL INDIO (ARGENTINA)

Alejandro García CONICET—Departamento de Historia, Facultad de Filosofía, Humanidades y Artes, Universidad Nacional de San Juan, Ignacio de la Roza 230 (Oeste)—5400 San Juan, Argentina. Email: [email protected] or [email protected].

ABSTRACT. New excavations and new 14C dates at the Gruta del Indio shelter in the central Argentinean Andes show that the dung layer of the site is much thicker towards the front of the site than near the rock wall. This yields a longer chronology for the dung deposit; thus, the upper boundary would date to about 9000 14C yr BP. The new measurements lengthen the possible time of the coexistence of man and extinct Pleistocene megafauna in the area, since ~1400 cal yr is much longer than previously thought (Long et al. 1998). Nevertheless, coexistence does not imply interaction, which still is not evident.

INTRODUCTION Gruta del Indio (34°45′S, 68°22′W) is a significant point of reference for the discussion of the features of the early human peopling of the central Argentinean Andes and the human relation to the extinction of Pleistocene megafauna (Long and Martin 1973; Lagiglia 1974; García 1999) and to the environmental changes (D’Antoni 1983; Dacar et al. 2001). Despite the finding of lithic artifacts in the lower stratigraphic levels and of hearths dated to the final Pleistocene (Lagiglia 1956, 1968, 1979), a recent analysis of 14C dates on charcoal and dung samples (Long et al. 1998) from the right area of the shelter (Figure 1) suggests that the coexistence of man and extinct sloths is not certain and, at most, it would have been very short. In Long et al. (1998), a set of 10 dates from dung samples put these materials between 9650 ± 800 and 24,730 ± 860 14C yr BP. The calibration (2σ) done by Long et al. (1998:696) for 5 dates with 1σ counting errors less than 200 yr indicates maximal ages ranging from about 12,000 to 14,680 cal yr BP. Another set of 10 dates was obtained from charcoal samples with 1σ counting errors less than 200 yr. Results span from 8045 ± 55 to 10,530 ± 140 14C yr BP, with calibrated ages (2σ) between 8600 and 12,980 yr BP. The maximal possible overlap, based mainly on dates A–9493 (dung) and A– 1638 (charcoal), is of 880 or 930 cal yr, depending on which calibration curve is used, from Burr et al. 1998 (12,880–12,000 = 880) or from Hughen et al. 1998 (12,980–12,050 = 930). Otherwise, the probability range (2σ) of a date from dermal ossicles (GrN 5772) overlaps or pre- cedes those of 9 dates on charcoal (Long et al. 1998:Figures 4a and 4b). However, because of its origin, Long et al. (1998:698) consider the date on dermal ossicles to be anomalous. On the basis of this information, the coexistence of man and extinct megafauna would not be clear. Recent excavations, carried out between 1997 and 2000, at the right area of the shelter and near the rain-drop line corresponding to the boundary of R8 and R9 squares (Figure 2), allow for a much broader chronological superposition. This proposal is based on a new series of dates from dung and charcoal samples as well as on stratigraphic observations.

METHODOLOGY Previous work at the site pointed out the presence of a layer containing extinct megafauna dung, ranging from about 0.15–0.30 m thick, although in some areas it could reach 0.40 m (Semper and Lagiglia 1968:133; Lagiglia 1974). Since these excavations were mainly in an area adjacent to the wall of the shelter (Figure 2), some seemingly anomalous dates (Lagiglia 1974), and the difficulty to determine the chronology of the layer are explained by its lesser width there.

33 34 Alejandro García

Figure 1 General view of the new excavation at the right area of Gruta del Indio

VSQ I B UT RR R POÑNMLLLKJ HGF EDCHC A 1 1 2 3 Left area 4 2 5 6 3 7 8 9 Right area Talus 10 Approximate beginning 11 of talus 12

Water drop line

Main excavated 0 10 m areas

Figure 2 Plan of the site and main areas excavated, re-drawn from Lagiglia 1977: 1=Lagiglia/Semper; 2=Long/Lagiglia; 3=García/Lagiglia

In order to minimize disturbance of the site, the new excavation was carried out at a restricted area of some 2 m2 cleared by partially removing two great rocks with a tackle. The results revealed that the layer including the megafauna dung is thicker towards the front of the shelter, at least in the area of squares R8 and R9 (García and Lagiglia 1999). The thickness of the layer reaches 0.9 m, allowing a better chronological control of the deposition periods. This wider stratigraphic distribution of the dung made it possible to obtain new dates that offer a very different view of the chronology of the Coexistence of Man and Megafauna at Gruta del Indio (Argentina) 35 layer. The analyses were run by the Tritium and Radiocarbon Laboratory (Latyr) at the Universidad Nacional de La Plata and Beta Analytic, Inc. According to the results, a thin portion of that layer (0.10–0.30 m) would have been deposited during the Pleistocene–Holocene transition (Table 2, date #9), while the lower portion (about 0.60 m) would have laid between at least ~36,000 and ~24,000 yr BP (Table 2, #10–17). Taking into account the small size of the recently analyzed dung balls (~80 mm), the probable presence of Machrauquenia sp. and Hippidion sp. (besides Megatherium sp. and Mylodon sp., see Sem- per and Lagiglia 1962–1968) in the record (Miotti, personal communication), and the morphological and dimensional similarity to that of actual equines, it is possible that the dung balls do not correspond to ground sloths. For a better comparison of the results, all dates were calibrated using the Calib revision 4.3 program (Stuiver et al. 1998) and no subtractions to the conventional ages (in order to compensate eventual interhemispheric differences) were made (Figini 1999). To incorporate the date A–1351 to the comparison set, all dates with 1σ counting errors ≤ 210 yr were included.

RESULTS AND DISCUSSION A date from a dung sample corresponding to the more recent depositions (close to the upper boundary of the layer) gave an approximate age for the end of the megaherbivores occupation at the site. The analyzed material (a single, 64-g dung ball) was taken at a distance of about 0.10–0.20 m from that boundary (Figure 3). The result was 8990 ± 90 14C yr BP (LP 925). From the same level of LP 925, we recovered some 60 little fragments of bones presumably belonging to megafauna and a small (13 mm) flake of rhyolite, probably dispersed vertically by postdepositional disturbance agents (García and Lagiglia 1999b).

Loose yellowish brown slime

Yellowish brown fine sediment

8990±90

24140±510 Brown fine sediment with megafauna feces

28670±720

Dark sand

Non-excavated area Rocks 20 cm Figure 3 Pit wall at the R8/R9 boundary 36 Alejandro García

Moreover, new dates on charcoal samples from previous and present work at squares Ñ4, O5, Q7 and RR9 (Lagiglia, personal communication) have been recently obtained. Results range between 7430 ± 90 and 9700 ± 110 14C BP, with calibrated ages (2σ) between 8030–8390 and 10,690–11,260 BP. Tables 1 and 2 show the calibrated dates (2σ) only with 1σ counting errors less than 200 yr from Long et al. (1998) and the new dates mentioned above. Even if the date on dermal ossicles rejected by Long et al. (1998) is removed, a difference of about 1420 cal yr (11,780–10,360 = 1420 cal yr) is recorded between the more conservative possibility of the date—corresponding to the older charcoal sample (A–1638, Table 1, #2)—and the oldest possible age of the more recent one, corresponding to the younger dung sample (LP–925, Table 2, #9). Considering 1σ, the lesser possible difference between the calibrated ages for A–1638 (12,180–12,890) and LP–925 (9920–10,220) is about 1960 cal yr. Finally, the maximal possible difference between the cal ages for both samples (2σ) is 3160 cal yr (12,950–9790 cal yr). Taking into account the presence of more recent dung (because of its stratigraphic position, see Fig- ure 3), the difference between the ages for the oldest charcoal and the youngest dung could be even greater. Table 1 Dates with standard deviations ≤210 yr from Long et al. (1998:696) Nr Materiala Lab. nr Coll.b Origin Date Cal 2σ Stuiver et al. 1998 1 Ch GrN 5394 HAL RR9, 0.7 m 8045 ± 55 8720 (9005) 9230 2 Ch A-1638 AL F3, 2.2 m 10530 ±140 11780 (12429, 12455, 12361) 12950 3 Ch A-9486 AL P6–O5, 0.8 m 10135 ± 95 11260 (11696, 11723, 11726) 12340 4 Ch A-9489 AL O5 9905 ± 140 10890 (11240, 11246, 11255) 11950 5 Ch A-9491 AL O5 9770 ± 85 10810 (11186) 11330 6 Ch A-9492 AL Fogón 3 9825 ± 95/90 11110 (11202) 11550 7 Ch A-9495 AL O5 9890 ± 75 11170 (11232) 11550 8 Ch A-9496 AL O5 9990 ± 75 11200 (11303, 11315, 11339, 11396, 11519, 11539) 11940 9 Ch A-9497 AL O5 10195 ± 80 11360 (11767, 11806, 11927) 12360 10 Ch A-9498 AL O5 10170 ± 70 11360 (11755, 11818, 11909) 12340 11 D GrN 5558 HAL RR8, 1.1 m 10950 ± 60 12650 (12980) 13160 12 D A-1351 HAL Q7-7, 0.7–0.8 m 10610 ± 210 11760 (12656, 12728, 12804) 13120 13 D A-1371 HAL Q7-8, 0.8–0.9 m 11820 ± 180 13420 (13826) 15250 14 D A-9571 AL P6, >0.6m 12375 ± 115 14100 (14325) 15510 15 D A-9570 AL P6, >0.6m 11040 ± 130 12660 (13013) 13370 16 D A-9493 AL P6, >0.6m 10900 ± 185 12390 (12944) 13190 17 B GrN 5772 HAL RR-9–7, 0.7 m 9560 ± 90 10580 (10767, 10832, 10837, 10957, 11006, 11017, 11059) 11180 aCh=charcoal; D=dung; B=bone bColl.=Collector: AL=Austin Long; HAL= Humberto Lagiglia Coexistence of Man and Megafauna at Gruta del Indio (Argentina) 37

Table 2 Dates recently obtained for lower layer at Gruta del Indio Cal 2σ Stuiver et al. Nr Materiala Lab. nr Coll.b Origin Date 1998 1 Ch LP-860 HAL O5, 1 m 9590 ± 120 10560 (10794, 10795, 10809, 10823, 10860, 10940, 11068) 11230 2Ch LP-941 HAL RR9, 0.7 m 9580 ± 100 10580 (10790, 10802, 10805, 10827, 10851, 10941, 11065) 11200 3Ch LP-991 HAL Ñ4–1, 9510 ± 90 10510 (10740) 11160 0.38–0.48 m 4Ch LP-986 HAL O5–2, 9160 ± 90 10180 (10242, 10352, 0.72–0.9 m 10355) 10580 5Ch LP-876 HAL O5, 9700 ± 110 10690 (11165) 11260 0.8–0.9 m 6Ch LP-854 HAL Q7–5, 8920 ± 110 9600 (9977, 9986, 0.5–0.6 m 10151) 10240 7Ch LP-873 HAL Q7, 7430 ± 90 8030 (8195, 8296, 0.35–0.4 m 8278) 8390 8Ch LP-845 HAL O5, 0.8 m 7860 ± 90 8420 (8606, 8620, 8628) 9010 9 D LP-925 AG R8/9 8990 ± 90 9790 (10190) 10360 10 D LP-929 AG R8/9 30200 ± 800 — 11 D LP-918 AG R8/9 30800 ± 700 — 12 D LP-1072 AG R8/9 28670 ± 720 — 13 D LP-1075 AG R8/9 24140 ± 510 — 14 D Beta 152587 AG R9 >37610 — 15 D Beta 152588 AG R9 29530 ± 540 — 16 D Beta 152589 AG R9 36400 ± 200 — aCh=charcoal; D=dung; B=bone bColl.=collector: HAL=Humberto Lagiglia; AG=Alejandro García

Also, date LP–925 is younger than 9 other dates from charcoal samples given by Long et al. (1998: Table 1, #2–10), and younger than 5 of the new results obtained by Lagiglia (Table 2, # 1–5). According to the tables above and assuming that the age of the dung and the charcoal from the site constitutes an adequate instrument to solve the problem, the possible time of man and megafauna coexistence in the area of Gruta del Indio would be much greater than previously thought. This situation would not be rare given the available information for the early human peopling of the region, as indicated by the archaeological record of Agua de la Cueva rockshelter (200 km north) with an initial occupation dated to 10,950 ± 190 14C yr BP (García et al. 1999). Nevertheless, this is not a proof for the interaction between man and megafauna. On the contrary, the signals are scarce (Semper and Lagiglia 1962–1968; Lagiglia 1979) and eventually suggest a very sporadic presence of man at the site during the Pleistocene–Holocene transition. Otherwise, although attention has been focused on the overlap of dates from dung and charcoal samples, better proof is still needed for the cultural origin of that charcoal. The absence of burnt dung in the area of R8/9 suggests that no natural fires expanding throughout the site would have occurred, so the charcoal recorded could have a cultural origin. Notwithstanding, recent observations point out marked 38 Alejandro García

Cal years BP Dung & bone Charcoal 13500

13000

12500

12000

11500

11000

10500

10000

9500

Figure 4 Overlap of probability ranges of calibrated dates (2σ) stratigraphic differences at neighboring areas of the site (García 2001), whereas the sole presence of definite charcoal patches or lines does not imply a cultural origin (García and Zárate 1999). Due to the scarcity of lithic artifacts (Lagiglia 1979) and their lack of direct association with the charcoal and megafauna items (Borrero 2002), the most fruitful way to prove the interaction of man and Pleistocene megafauna (not considering new excavations in the central area of the site) seems to be the study of the osseous material showing probable cultural marks, discovered at the right area of the shelter (Semper and Lagiglia 1962–1968; García and Lagiglia 1999).

CONCLUSIONS The 14C dates from Gruta del Indio indicate that several episodes of charcoal burning occurred during at least the last 1400 cal yr of existence of the Pleistocene megafauna around the site. If the cultural character of that charcoal is accepted (Long et al. 1998) and an effective interaction is assumed (Lagiglia 1974), the time of coexistence between man and Pleistocene megafauna in the area could be so prolonged as to allow interaction and the incidence of the former to the extinction. However, to prove that interaction, firmer, more solid evidence is still required, which could come from the future discovery of cultural artifacts and megafauna items in direct association or from the analysis of use-traces and fractures on the archaeofaunistic material presently available.

ACKNOWLEDGMENTS To Humberto Lagiglia, for his suggestions and help in the fieldwork, and to the owners of the Campo Limeño for permitting us the access to the site. I also wish to thank Paul Martin and Austin Long for their helpful comments on the manuscript. Recent research at Gruta del Indio was supported by Fundación Antorchas. Coexistence of Man and Megafauna at Gruta del Indio (Argentina) 39

REFERENCES Borrero LA. 2002. Arqueología y biogeografía humana García A, Zárate M. 1999. Perdurabilidad y cambios de en el sur de Mendoza. In: Gil A, Neme G, editors. En- fogones experimentales en la precordillera men- tre montañas y desiertos: arqueología del sur de Men- docina. Arqueología 9:105–21. doza. p 195–202. García A, Zárate M, Páez M. 1999. The Pleistocene–Ho- Burr GS, Beck JW, Taylor FW, Recy J, Edwards RL, Ca- locene transition and human occupation in the Central bioch G, Corrège T, Donahue DJ, O’Malley JM. 1998. Andes of Argentina: Agua de la Cueva locality. Qua- A high-resolution radiocarbon calibration between ternary International 53/54:43–52. 11,700 and 12,400 calendar years BP derived from Hughen KA, Overpeck JT, Lehman SJ, Kashgarian M, 230Th ages of corals from Espiritu Santo Island, Van- Southon J, Peterson LC, Alley R, Sigman DM. 1998. uatu. Radiocarbon 40(3):1093–105. Deglacial changes in ocean circulation from an ex- Dacar M, Borghi C, Martínez Carretero E, Giannoni S, tended radiocarbon calibration. Nature 391:65–8. García A. 2001. Paleodiet of Cavidae rodents in the Lagiglia H. 1956. Estudios Arqueológicos en el Rincón Monte Desert of Argentina in the last 30,000 years. del Atuel (Departamento San Rafael, Mendoza). Current Research in the Pleistocene 18:95–7. Anales de Arqueología y Etnología XII:227–87. D’Antoni H. 1983. Pollen analysis of Gruta del Indio. Lagiglia H. 1974. Atuel IV frente a la prehistoria ameri- Quaternary of South America and Antarctica Penin- cana. Paper presented at III Congreso Nacional de Ar- sula. p 83–103. queología Argentina. Salta. Figini A. 1999. Análisis de la calibración en años calen- Lagiglia H. 1977. Arqueología y ambiente natural de los darios de las edades C-14. Corrección para el Hemis- valles del Atuel y Diamante. PhD dissertation. Facul- ferio sur. In: Actas del XII Congreso Nacional de Ar- tad de Ciencas Naturales y Museo, UNLP. queología Argentina II:349–52. Lagiglia H. 1979. Dinámica cultural en el centro oeste y García A. 1999. La extinción de la megafauna pleis- sus relaciones con áreas aledañas argentinas y chile- tocénica en los Andes Centrales Argentino-Chilenos. nas. In: Actas del VII Congreso de Arqueología de Revista Española de Antropología Americana 29:9– Chile II:531–60. 30. Long A, Martin P. 1973. Death of American Ground García A. 2001. Estado actual de los estudios sobre el po- Sloths. Science 186:638–40. blamiento temprano del centro de Mendoza. Paper Long A, Martin P, Lagiglia H. 1998. Ground Sloths and presented at XIV Congreso Nacional de Arqueología humans at Gruta del Indio–Argentina. Radiocarbon Argentina. 40(2):693–700. García A, Lagiglia H. 1999a. A 30,000-year-old Mega- Semper J, Lagiglia H. 1962–1968. Excavaciones ar- fauna dung layer from Gruta del Indio (Mendoza, Ar- queológicas en el Rincón del Atuel (Gruta del Indio). gentina). Current Research in the Pleistocene 16:116– Revista Científica de Investigaciones I(4):89–158. 18. Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, García A. 1999b. Avances en el estudio del registro pleis- Hughen KA, Kromer B, McCormac FG, van der Plicht tocénico tardío de la Gruta del Indio (Mendoza). In: J, Spurk M. 1998. INTCAL 98 radiocarbon age cali- Cuadernos del Instituto Nacional de Antropología y bration, 24,000–0 cal BP. Radiocarbon 40(3):1041– Pensamiento Latinoamericano 18:167–74. 83. RADIOCARBON, Vol 45, Nr 1, 2003, p 41–58 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

EL MIRÓN CAVE AND THE 14C CHRONOLOGY OF CANTABRIAN SPAIN

Lawrence Guy Straus Department of Anthropology, University of New Mexico, Albuquerque, New Mexico 87131, USA. Corresponding author. Email: [email protected]. Manuel Gonz·lez Morales Departamento de Ciencias Históricas, Universidad de Cantabria, 39005 Santander, Spain. Email: [email protected].

ABSTRACT. Excavations since 1996 in the large El Mirón Cave in the Cantabrian Cordillera of northern Spain have revealed a cultural sequence of late Mousterian, early Upper Paleolithic, Solutrean, Magdalenian, Azilian, Mesolithic, Neolithic, Chalcolithic, Bronze Age, and Medieval occupations. These components have been dated by 51 generally coherent radiocarbon determinations, all run by the Geochron labs, in association with the Lawrence Livermore labs for AMS. This series is one of the largest for a single prehistoric site in Iberia or even Europe. The series is consistent with the record from Cantabrian Spain and provides new detail on the age of the Middle–Upper Paleolithic transition, on the various phases of the Magdalenian culture, on the appearance of the Neolithic in the Atlantic zone of Spain, and on the origins of the socioeconomic complexity in the metal ages. The stratigraphic relationship of 14C-dated levels to a roof-fall block and adjacent cave walls (both with engravings) provides rare terminus post and ante quem ages for execution of the rupestral art in El Mirón during the early to mid Magdalenian. The 14C record has also been instrumental in revealing the existence of depositional hiati during the early Holocene.

INTRODUCTION To avoid the pitfalls of using multiple laboratories, current prehistoric research in the Río Asón Val- ley in Cantabria, Spain (including not only the El Mirón Cave excavation co-directed by the present authors with 51 radiocarbon determinations, but also those of La Fragua, El Valle, and El Horno caves, directed by Gonz·lez Morales, M P García-Gelabert, and M Fano, respectively) has exclusively used the services of Geochron for 14C dating (Gonz·lez Morales and Straus 2000; Straus et al. 2001, 2002 a,b). The present article is, therefore, a posthumous homage to Hal Krueger, a pioneer in archeological isotopic analyses and in high-quality professional service to the disciplines of archeology and geology. It was Hal who ran the dates from the first season at El Mirón in 1996 with results as promising as they were spectacular. Alex Cherkinsky, who ran most of the dates after Hal, has continued Krueger’s tradition of rigor and rapidity at Geochron, which are part of Hal’s legacies.

14C DATING THE STONE AGE PREHISTORY OF CANTABRIAN SPAIN The first attempts at 14C dating Paleolithic materials (namely, charcoal from the Magdalenian sites of Altamira and El Juyo and, less successfully, mollusc shells from Altamira) in the Cantabrian Region of Spain were made by H R Crane and J B Griffin (1960) of the University of Michigan over 40 yr ago. The results of the charcoal dates have been generally confirmed with determinations from new excavations in both sites (Freeman 1988, 1996). The first excavation project in this classic prehistoric culture area to incorporate 14C dating as an integral part of its modern, interdisciplinary methodology was that of Cueva Morín, directed by J Gonz·lez Echegaray and L G Freeman in the late 1960s. The dating was done entirely by R Stuckenrath (1978) at the now-defunct Smithsonian Institution Radiation Biology Laboratory. Despite some stratigraphic inversions and a high degree of imprecision among some of the determinations, the 9 dates from MorÌn are still cited as among the relatively few that we have for the initial phases of the Upper Paleolithic in Cantabria. Following on the MorÌn Project one of its student participants, G A Clark (1976), undertook to definitively resolve the question of the age of the “Asturian culture”, in large part by the first-ever application of 14C dating to samples from the shell middens of eastern Asturias.

41 42 L G Straus, M González Morales

Figure 1 Map of the Asón River valley in eastern Cantabria: El Mirón & El Horno caves are two of the sites at 11; El Perro & La Fragua at 1; El Valle at 6; Cubio Redondo at 13; Tarrerón at 12

When organizing the La Riera Paleoecological Project in 1975, Clark and Straus proposed to make extensive use of 14C dating to provide the chronological framework of the sequence at La Riera Cave in eastern Asturias, rather than essentially relying on “diagnostic” artifact presence or percent- ages to attribute specific strata to particular cultural-historical phases. The result of systematic 14C assays for the 2.5 m of deposit (30 levels) was a list of 28 dates ranging in age from about 21,000 to 6500 BP (Straus and Clark 1986). Unfortunately, in part to take advantage of some offers of free dates, we obtained determinations from 5 different laboratories. Due to this fact and the subsequent interlaboratory errors (International Study Group 1982), combined with undoubted interlevel disturbances and sample movements in a stratigraphy composed of very thin levels with no culturally sterile zones, it is clear that there are many reasons probably responsible for some of the incoherences El Mirón Cave & the 14C Chronology of Cantabrian Spain 43

Figure 2 Map of the central part of Cantabrian Spain, showing locations of the Asón River drainage and of some major Paleolithic sites, 2=La Riera; 3= Altamira; 4=El Juyo; 5=Morín; 6=Castillo; 7=Rascaño among the La Riera dates, as have often been commented upon and criticized. Nonetheless, the La Riera sequence, in association with other sites that were being 14C-dated in the 1970s and 80s, clearly established the following sequence for the late Upper Paleolithic and Mesolithic in the Cant- abrian region: • Solutrean: 20,500–17,000 BP; • early (= Lower + Middle) Magdalenian: 17,000–13,000 BP; • Upper Magdalenian: 13,000–11,500 BP; • Azilian: 11,500–9000 BP; • Asturian/Mesolithic: 9000–6000 BP (all dates uncalibrated). Subsequently, several other excavation projects in northern Atlantic Spain have invested fairly heavily in 14C dating despite initial, and some on-going, skepticism about the reliability of the method on the part of some traditional prehistorians. Notable in this context have been the Middle Río Nalón Project organized by J Fortea (and in which Gonz·lez Morales and M S Corchón have been major participants), numerous excavations in the Basque province of Guip˙zcoa directed by J Altuna et al. of the Aranzadi Scientific Society (e.g., Ekain, Erralla, Amalda, Aitzbitarte, Labeko), the Río Asón Estuary Project directed by Gonz·lez Morales (1995), the re-excavation of El Castillo Cave directed by V Cabrera (Cabrera and Bernaldo de Quirós 2000; Cabrera et al. 1996), the La Garma Complex and Cantabrian Neolithic Projects directed by P Arias (Arias et al. 1999, 2000), and most spectacularly, the various projects to directly date Cantabrian cave art by AMS, the latest 44 L G Straus, M González Morales results of which have recently been reported by Moure and Gonz·lez Sainz (2000 a,b) and Fortea (2002). There is no up-to-date, global compilation of 14C dates for the Upper Paleolithic and Mesolithic of northern Spain. However, there is a complete list of all 14C dates (including even the Iron Age, Roman period, and Middle Ages) from the Spanish (and French) Basque country as of a decade ago (Mariezkurrena 1990). Straus (1992) published all known 14C dates for the Paleolithic and Mesolithic of the whole geographic macro-region (the autonomous administrative regions of Euskadi, Navarra, Cantabria, and Asturias). There have been no major compilations since then, except the Alvarez and Jˆris’ (1998) calibration of dates for the mid-Magdalenian at Las Caldas and other sites.

Figure 3 Plan of the El Mirón Cave vestibule, showing the location of our excavations

EL MIRÓN CAVE El MirÛn is a large cave in Monte Pando. It was scientifically discovered by L Sierra and H Alcalde del Río in 1903 (along with the adjacent cave art sites of Covalanas and La Haza), but was largely ignored or written off as totally disturbed ever since. This massif of highly-karstic, Lower Creta- ceous limestone is one of the coastward ranges of the Cantabrian Cordillera in easternmost Cant- abria Province, on the border with Vizcaya and near the border with the meseta of Burgos (Old Castile) at the low (920 m above sea level) Los Tornos Pass (Figure 1). The site is at the strategic confluence of 2 tributary gorges with the major Ruesga Valley (the intermontane course of the RÌo Asón) a river that drains a large region midway between the coastal cities of Santander and Bilbao (Figure 2). Excavation of El MirÛn represents the first large-scale, modern-quality, archeological research project to be conducted in the mountainous interior of Cantabria. Until recently, most excavations had been done at sites on or near the narrow coastal plain along the Bay of Biscay. The large, flat-floored cave vestibule is mainly dry and faces due west. Located above the important market town of Ramales on the valley floor, El MirÛn was occupied by humans and livestock until recently. It is surrounded by about a dozen known cave art localities, none excavated. Our excavations in El MirÛn (seven 2-month campaigns since 1996) have been concentrated in 2 areas of the large (10 × 30 m) vestibule: front (“Cabin”) and rear (“Corral”), each at most about 10 El Mirón Cave & the 14C Chronology of Cantabrian Spain 45 m2 in size (Figure 3). We have connected these 2 block excavations with a continuous stratigraphic trench (8 m long × 1 m wide). We also regularized and deepened a never-published trench that had been dug in the dark inner cave during the 1950s. Access to the lower part of the stratigraphy was provided by emptying and screening the totally-mixed contents of a large pothole at the foot of a steep slope of (Tertiary?) colluvial-alluvial sediments leading up into the inner cave at the back of the vestibule. The clandestine digging had stopped at the base of the series of organically-rich Magdalenian layers, leaving intact all strata from Solutrean times downward. In the vestibule front area, the cultural-historical sequence currently includes layers pertaining to the Bronze Age, Chalcolithic, Neolithic, Mesolithic, Azilian, and the Upper and Lower/Middle Magdalenian (Figure 4). The main excavation in the former corral lacks the ceramic components

Figure 4 Stratigraphic section of the excavation near the front of the vestibule (“Cabin” area) 46 L G Straus, M González Morales

Figure 5 Stratigraphic section of the excavation at the rear of the vestibule (“Corral” area) and has a sequence of possible Mesolithic, then Azilian, Upper, Middle, and Lower Magdalenian layers (Figure 5). The 2 × l m test pit in the base of the pothole has a sequence of Solutrean, Early Upper Paleolithic, and Mousterian strata (Figure 6). To date, the mid-vestibule connecting trench has revealed remnant Bronze Age/Chalcolithic and Neolithic deposits toward the cave mouth; these are underlain by a nearly sterile layer of Mesolithic age and then Azilian and Upper and Middle Magdalenian levels (Figure 7). The inner cave trench revealed a bonfire layer of Medieval age with only faunal remains, a series of sterile clay and mondmilch layers (but with possible torch fragments dating to the Bronze Age), and at the base, a Middle Magdalenian layer (Figure 8). Ground-penetrating radar, magnetometry, and electrical resistivity surveys of the vestibule have shown that there are approximately 9 m of sedimentary deposits above the bedrock. In aggregate total at the rear of the vestibule, we have excavated to a depth of about 5 m below the ground surface, and in the front, we have dug down no more than about 3.5 m as of the end of the summer 2002 campaign, although core boring in 2003 showed at least approximately 1 m more cultural deposits below the base of the excavation. El Mirón Cave & the 14C Chronology of Cantabrian Spain 47 e Calibrated date d alenian; LM=Lower Magdale- etween Middle & Lower Magdalenian are etween Middle & Lower Lab nr Method c ilian & Upper Magdalenian and b =Azilian; UM=Upper Magdalenian; MM=Middle Magd Date (BP) 1 SD Material rral”); MV=mid-vestibule (“Trench”) olithic (Distinctions between Az b =Chalcolithic; Neo=Neolithic; Mes=Mesolithic; AZ (“Cabin”); VR=vestibule rear (“Co nal; Cxcnt=extended count Tc=tooth collagen; Bc=bone collagen Tc=tooth C dates (1996–2002) Level Spit Period 14 B 4.1.2 (Range at 1 sigma) a tentative) nian; Sol=Solutrean; EUP=Early Upper Paleolithic; MP=Middle Pale Med=Medieval; BA=Bronze Age; Chal AMS=accelerator; Conv=conventio IC=inner cave; OV=outer vestibule bioapatite; Ch=charcoal; Ta=tooth et al. (1998); CALI Stuiver H2J2I3H2 OVI3 OVJ2 OVH4 OV 3I4 OV 5I3 OVI3 5.1 OV 7 4J4 OVI4 8.1 3 OV 4 9I4 OV 14 9I3 13 OV BAJ2 9.6 OV 10 ChalI3 Chal 8 OV 22 10 ChalI3 22 OV 10.1 Neo?I3 19 OV 10.1 3700J3 19 OV 10.1 34 Neo 3820 NeoJ3 4120 OV 11.1 Neo 28 3740J2 OV 4680 12 29 NeoO6 OV 40 15 25 Neo 240L5 Mes OV 15 50 5170L5 Ch 5280 120 Mes 12 OV 16 Ch 5250L5 MV 60 Mes 43 Ch 17 Ch 5570 MV AZ/UM 43 17 Ch GX-25851 5690 170 MV 44 GX-22127 8380 17 AMS 302? UM 40 Conv 150 MV GX-22130 32 8700 11,720 GX-22460 Ch 303 LM AMS 39 Cxcnt 9550 Ch GX-22131 50 Ch 303.1 LM 2140–2030 BC Pit98a 20 AMS 2575–1931 BC 175 50 303.3 LM 2858–2586 BC Ch 12,970 2305–1963 BC GX-22128 13 140 LM 14 40 GX-24461 Conv Ch Neo? Ch GX-24462 3612–3371 BC 15,010 LM 16 AMS 50 Cxcnt Ch 15,220 Ch LM GX-23414 4221–3789 BC 15,180 Ch AMS GX-24463 Neo 70 GX-23413 4217–4001 BC 4318–3945 BC Neo 15,470 Cxcnt 260 AMS GX-23391 4910 GX-25852 Neo 15,450 Bc 300 Conv 4449–4359 BC AMS GX-24464 Ch 15,700 100 7586–7182 BC 4582–4458 BC AMS Bc BC 12,039–11,523 240 5500 7745–7609 BC Bc GX-22132 5520 160 80 GX-23392 AMS BC Bc 9119–8792 5790 190 Conv GX-23393 Bc Ch BC 13,990–13,299 AMS GX-23415 Ch BC 16,396–15,609 90 Conv GX-24466 70 16,679–15,810 BC Conv GX-28211 GX-27115 90 Ch BC 16,487–15,910 Conv Cxcnt Ch GX-25853 BC 16,919–16,145 Conv Ch 16,852–16,184 BC 3773–3641 BC GX-25854 BC 17,151–16,441 GX-25855 Conv Conv GX-25856 Cxcnt 4451–4250 BC 4451–4261 BC 4768–4540 BC Square Zone Table 1 Table El Mirón Cave a b c d e 48 L G Straus, M González Morales e alenian; LM=Lower Calibrated date d n and between Middle & Lower dirt and dung Lab nr Method Magdalenian; MM=Middle Magd c n & Upper Magdalenia h GX-27521c AMS 10,665–10,024 BC top ~20 cm of recent 0 Ta GX-27521a AMS 10,984–10,699 BC ctions between Azilia after removal of the lithic; AZ=Azilian; UM=Upper Date (BP) 1 SD Material rral”); MV=mid-vestibule (“Trench”) b (Continued) Med 540 100 Ch GX-24465 Cxcnt AD 1304–1442 Paleolithic; MP=Middle Paleolithic (Distin lithic; Neo=Neolithic; Mes=Meso from the indicated stratum f 8 MM 14,850 60 Ch GX-27114 AMS 16,092–15,559 BC 26 LM 17,400 80 Ch GX-29439 AMS 19,082–18,424 BC to the excavation from Corral area surface =tooth collagen; Bc=bone collagen (“Cabin”); VR=vestibule rear (“Co ” g tional; Cxcnt=extended count C dates (1996–2002) Level Spit Period 14 hearth pit dug within or a ); CALIB 4.1.2 (Range at 1 sigma) Magdalenian; Sol=Solutrean; EUP=Early Upper Magdalenian are tentative) Med=Medieval; BA=Bronze Age; Chal=Chalco AMS=accelerator; Conv=conven Sample charcoal is from IC=inner cave; OV=outer vestibule bioapatite; Tc Ch=charcoal; Ta=tooth et al. (1998 Stuiver Large lump of charcoal apparently fallen in Large V8V8W10W10 VRX10 VRAZICFlowstone VRX10 VR 117 Slope top VR 119 —Slope 125 VR8-9Q 126 2411Q top IC 128 2811R IC 4 130 Flowstone AZ 10,740 5 LM IC 20 LM/Sol IC 36 Sol IV — Sol 4 VII 16,960 EUP 17,050 VIII MP — 18,980 — 10,390 18,950 10 27,580 80 60 Med 41,280 Ch 360 BA Ch MM 350 210 50 Bc 1120 GX-25858 Bc GX-25857 Ch C AMS AMS 900 Ch GX-24470 14,620 3230 18,570–17,923 BC Conv 18,760–18,031 BC GX-24471 GX-27113 Conv AMS GX-27112 21,106–20,035 BC 80 AMS 21,065–20,007 BC 80 40 Ch Ch Ch — GX-22129 — Conv GX-22347 GX-28013 AMS AMS AD 1040–1204 15,818–15,290 BC 1523–1441 BC P6P6P6U7 MV MV MV 305 VR 306 308 surface 9 11 16 # AZ AZ/UM UM 11,650 10,270 12,350 50 50 Bc 180 Bc Ch GX-24468 AMS GX-24467 AMS GX-28210 BC 11,861–11,525 Cxcnt 10,362–9818 BC 13,403–12,166 BC T8V8V8T10 VR VR VR VR 102.1 108 108 3 108 4 5 AZ/UM MM 11,950 MM 13,660 70 14,710 Bc 70 160 GX-23417 Bc Bc AMS BC 12,136–11,890 GX-22703 GX-23397 AMS Conv 14,689–14,209 BC 15,969–15,346 BC V8V8U7T10 VRV8 VRV8 VR VR 110V7 111 VR 111 VR 114 8 VR 14 115 20 116 17 116 MM MM 19 MM 20 LM 16,130 LM 16,370 LM 15,530 16,460 250 13,800 190 15,220 230 Bc Bc 50 Bc 840 GX-23396 Bc 100 GX-23395 Conv Bc Conv GX-24469 Bc Conv 17,699–16,884 BC GX-28209 17,935–17,201 BC AMS GX-23394 16,981–16,221 BC Conv GX-23416 17,989–17,364 BC AMS 15,725–13,517 BC 16,534–15,955 BC Square Zone Table 1 Table El Mirón Cave a b c d e f g El Mirón Cave & the 14C Chronology of Cantabrian Spain 49

Figure 6 Stratigraphic section of the deep sondage at the vestibule rear

To avoid some of the problems that may have arisen by using many different 14C laboratories at La Riera Cave, we decided to run all El MirÛn dates through the Geochron lab, based on its excellent track record in providing dates for the South Belgium Prehistoric Project directed by Straus and M Otte. Up until now, 51 determinations from El Mirón Cave have been done, to our knowledge mak- ing it the largest series of 14C dates from any single prehistoric site in Spain (see Table 1). The determinations have been done on both charcoal and bone collagen, conventionally or by AMS. Calibra- tions have been run for dates back to about 20 kya using the latest available version of CALIB (Stuiver et al. 1998).

14C CHRONOLOGY OF THE POST-PLEISTOCENE LEVELS Discussion of post-Pleistocene ages is in terms of calibrated AD and BC (Figure 9). There are 2 14C dates of Medieval age: 1 from a charcoal-rich layer (Level IV) stretching from wall-to-wall across the inner cave. The other is from a large lump of charcoal that came from the ground surface at the rear of the vestibule, after modern dung and debris had been shovelled out. The inner cave date (11th–12th century AD) might refer to a period of insecurity in the region, when other indicators suggest that people often took refuge and cached valuables in caves. The vestibule date (14th–15th 50 L G Straus, M González Morales

Figure 7 Stratigraphic section of the central part of the mid-vestibule connecting trench century AD) attests to the continuing use of the cave by people and animals, as do the presence of the well-built stone cabin foundations in the outer part of the El Mirón vestibule. The topmost, major prehistoric cultural level in the Cabin area (Stratum 3) is very rich in ceramics of Bronze Age type, domesticated animal remains (especially cattle and pig, as well as sheep/goat), and slag. It also yielded a copper pin and abundant evidence of in situ combustion, probably related (at least in part) to metallurgy. The date of ~2100 cal BC is consistent with an early Bronze Age, which, at any rate, is poorly distinguished from the Chalcolithic in this then culturally-peripheral region of the Iberian Peninsula. The succession of massive ash and charcoal lenses (possibly represented episodes of hygienic straw- and dung-burning) and pits of various sizes, contents, and probably functions that make up Strata 4–7 testify to intensive human occupation and animal stabling in El Mirón Cave & the 14C Chronology of Cantabrian Spain 51

Figure 8 Stratigraphic section of the inner cave trench (not shown in Figure 3) the El Mirón vestibule during Chalcolithic times. The temporally diagnostic stone arrowheads and ceramics are validated by dates centered on about 2500 cal BC. The abundance of pitting in these levels probably explains the stratigraphic inversion of GX-22130 and 24460. Again, the significance and timing of the transition between the Neolithic and Chalcolithic in the Cantabrian region are still poorly defined (hence, the significance of such modem excavations as those at El MirÛn or Pareko Landa in nearby central Vizcaya [Aguirre et al. 2000]). The Neolithic range of levels (10–8 and 303.3–303) at El Mirón dates between about 4600 and 3500 cal BC. How- ever, the top of the sequence is currently undated, probably leaving a gap of about 500 yr precisely at the time of the transition, although this might be partly filled by a date of about 3700 BC (GX- 28211) on abundant charcoal from the base of a large pit possibly corresponding to Level 302. (The top of this crater-like, 50-cm deep feature may still have been in use in the Bronze Age, based on the presence of possible slag near its surface.) The basal Neolithic dates from El Mirón (GX-25856 and 23413) are among the oldest that are known from the northern slope of the Cantabrian Cordillera (Arias et al. 1999) and are particularly important as they are definitely associated with domesticated ovicaprines (J Altuna and K Mariezkurrena, personal communications). They also correspond to the very oldest dates for the construction of megalithic monuments in the region (Yarritu and Gorro- chategui 1995; Serna 2000). Although there is no direct palynological evidence for agriculture at El MirÛn, wheat grains first appear in Level 303 at about 4300 cal BC (MJ Iriarte and L Peña, personal communications). Cereal grain has also been positively identified at Kobaederra Cave in central Vizcaya and is directly dated to about 4200 cal BC (Aguirre et al. 2000). Other sites 14C dated to about the same age lack ceramics and domesticated animals, but have double-bevel retouched lithic segments characteristic of the early Neolithic, with or without imported cereals [e.g., Herriko Barra in coastal Guipúzcoa; Pico Ramos Cave in coastal Vizcaya; Tarreón and Cubio Redondo caves— both the latter close to El Mirón [Mariezkurrena and Altuna 1995; Zapata 1995; Apell·niz 1971; Ruiz and Smith 2001]). 52 L G Straus, M González Morales

Figure 9 14C dates for the post-Paleolithic strata of El Mirón Cave (with 1 sigma)

There are depositional hiati between Strata 10.1 and 10 and between 11 and 10.1 in the Cabin area; similar gaps probably also exist in the mid-vestibule connecting trench. If the 3 14C dates are all correct, the 8–9-cm-thick Stratum 10.1 formed very slowly during the millennium between about 8900–7500 cal BC. There are only small numbers of culturally non-diagnostic lithic artifacts and faunal remains (none domesticated) in this layer and the same can be said for probably contempora- neous Strata 304 and 101/102 in the mid and rear vestibule, respectively. These scarce finds attest to ephemeral visits to the cave during early Mesolithic times, when human settlement was concentrated along the coast, notably at the Río Asón estuary (González Morales 1995; Straus and González Morales 2003). Other visits to the cave occurred in late Azilian times, at the Pleistocene– Holocene boundary, as attested by 2 AMS dates of about 10,700 cal BC on the same tooth from a breccia remnant adhering to the cave wall at the top of the inner cave slope.

14C CHRONOLOGY OF THE PLENI- AND TARDI-GLACIAL LEVELS Over half of the dates (29) from El Mirón cover the periods of the Solutrean, Magdalenian, and Azil- ian, which together make up the late Upper Paleolithic (Figure 10). Dates for the Upper Paleolithic El Mirón Cave & the 14C Chronology of Cantabrian Spain 53 are reported in terms of uncalibrated BP determinations, although calibrated ages are given in Table 1. The distinction between the final Magdalenian and the Azilian is fairly arbitrary in the absence of the characteristic flat-section Azilian harpoons at El Mirón, since small-backed stone points and thumbnail endscrapers (present at the site) can be found in both periods. There is no single and absolute date at which the transition occurs and indeed, even within a geographic region such as Cant- abrian Spain, it can be time-transgressive between the Allerˆd and Dryas III climatic phases. Stra- tum 305 in the mid-vestibule trench gave a date that is of late Azilian age (10,270 BP) but then there is a series of dates ranging between 11,650–11,950 BP for Strata 11.1, 306, and 102.1 in the outer, mid, and rear parts of the vestibule, respectively. A red ochre-stained pebble in 11.1 could be considered an Azilian diagnostic. In contrast, there is great clarity concerning a charcoal date of 12,970 BP very closely associated with a fragment of a circular-section, unilaterally-barbed harpoon diagnostic of the Upper Magdalenian in Cabin Stratum 12. (There is a date of 12,960 ± 50 [GX-29440] from Level 4 of La Fragua Cave at the present mouth of the Asón.) Strata 103–107 in the Corral area, although not yet dated by 14C, probably pertain to this period. In general, however, the Upper Magdalenian is rather poorly represented at El Mirón but it and the Azilian have very abundant cultural remains (including harpoons pertaining to both periods) in Horno Cave at the base of the same mountain where Mirón is located and in El Valle Cave, about 8 km downstream along the Asón. These 2 sites have a total of 8 GX dates ranging between about 13,800–10,100 BP with 6 of them lying between 11,000–12,000 BP (Straus et al. 2002 a,b).

Figure 10 14C dates for the Late Upper Paleolithic strata of El Mirón Cave. 54 L G Straus, M González Morales

The culturally-richest Paleolithic layers excavated so far at El Mirón pertain to the early–middle Magdalenian period. This period—variously subdivided into phases by different specialists—is also very well and richly represented in such classic sites as Altamira, El Castillo, El Rascaño, and El Juyo in central Cantabria, as well as at many sites in Asturias, Guipúzcoa, and Navarra. It seems to have been a time of high population density with many sites and frequent use of favored caves both for residence and for decoration. Over 40 directly AMS-dated rupestral paintings from Asturias and Cantabria (including all the dates from Altamira) pertain to the period between 17,000–13,000 BP uncalibrated, and most all of those actually date around 14,000–15,000 BP (Moure and González Sainz 2000 a,b; Fortea 2002). Engravings on a huge block that had fallen from the El Mirón ceiling at the vestibule rear were executed at this time, since the block fell atop a level (110) dating to about 16,000 BP, after which its inner face was engraved and finally covered over by a series of later Magdalenian levels (109–103) during a period of some 4000 yr (González Morales and Straus 2000). In addition, engravings (including one of a horse) on the cave walls adjacent to the Corral excavation area at the vestibule rear can also be reasonably attributed to the mid-Magdalenian period, based on their position (roughly equivalent to the height of an adult human arm) relative to 14C-dated levels (García 2001). The 14C record for the early–mid Magdalenian in the Cabin excavation is straightforward and coherent when single standard deviations are considered: 6 determinations for Strata 15–17 lie between 15,000–15,700 BP. The situation is more complex in the Corral area. One determination (GX- 23394: a conventional date on bone collagen from Stratum 115) is clearly “too young” in comparison to dates stratigraphically above and below it. However, with an exceptionally large, single standard deviation of 840 yr it could be as old as 15,480 BP at +2σ, which is more in line with the general trend for Strata 110–116. Organically and artifactually rich, Stratum 108 has 3 determinations from a single m2 that are in stratigraphically descending order from 13,660–14,850 but the bone used for the topmost one could possibly have moved down from the rodent-burrow-ridden Stratum 107. While consideration of the standard deviations can eliminate some of the apparent inversions among the other dates from Strata 110–115, it is also true that all these levels are absolutely contig- uous with no intervening sterile zones. In fact, the subdivision of this whole dark brown, artifactually-rich deposit is a relatively subjective affair, based largely on variations in the relative amounts and size of limestone spall (éboulis) versus purer silt, which may be local in nature. Intensive human activity may have been responsible for the date inversions. All of these levels are very well- endowed in backed bladelets, antler points (sagaies), bone needles, and faunal remains (notably of ibex and salmonids) with abundant evidence of fire, including a well-preserved hearth-and-pit complex in Stratum 108. The latter is precisely dated by AMS on a chunk of conifer wood charcoal (identified by L Zapata) to 14,850 BP. It is in this period (corresponding roughly to Cabin Stratum 14 and Corral Stratum 108) that human use was made of the dark inner cave, where blade cores and blades have been found associated with a charcoal sample dated to 14,620 BP at the base of a test pit dug below the floor of the 1950s trench. Cabin Stratum 17 (also extremely densely littered with stone and bone artifacts and faunal remains, as well as evidence of fire) can be approximately correlated temporally by 14C with Corral Strata 110–116 in the 15,500–16,000 BP range. The seemingly “old” date of 17,400 BP (GX-29439) from Stratum 116 in V7b might be explainable by the fact that the charcoal sample comes from a hearth pit that might have been dug into or disturbed underlying layers. Approximate correlation of Cabin Strata 15/16 with Corral Stratum 108 is suggested by the discovery of the surface of a similarly distinctive, chocolate-brown, culturally-fertile deposit (312) at appropriate depths in a series of small test pits dug along the base of the mid-vestibule trench. El Mirón Cave & the 14C Chronology of Cantabrian Spain 55

The base of the Magdalenian sequence, which has more use of local, non-flint raw materials (especially to make “archaic-looking” macrolithic tools such as sidescrapers, denticulates, and notches) dates to about 17,000 BP (Strata 117 and 119). A couple of Solutrean point fragments were actually found in Stratum 119, indicating that the transition between the normatively defined Solutrean and early Magdalenian industries occurred at right about this time. This is completely in line with what has been shown at other sites in the region such as El Rascaño, La Riera, and Chufin (Straus 1992). The earlier parts of the chronostratigraphic sequence have so far only been revealed in the sondage dug below the bottom of the large pothole at the foot of the slope in the vestibule rear, not an ideal place for human habitation (due to wind currents) or preservation (due to erosion). Nonetheless, the results are encouraging as to the possibilities of finding Solutrean, early Upper Paleolithic, and Mousterian occupations elsewhere in El Mirón. The remnant Solutrean levels of 121–127) (partially cut into by the pothunters) are extraordinarily rich in foliate (including concave base), shouldered points, perforated shells and teeth, and fish remains, but have relatively few other artifacts. These layers, which are far less dark and organically rich than the overlying Magdalenian ones, are quite thin, so it is not surprising that Strata 125 and 126 have 2 statistically indistinguishable dates of 19,000 BP. This is a normal age for Solutrean occupations throughout Vasco-Cantabria and elsewhere in Iberia and southern France (Straus 1991, 2001) and corresponds to the onset of the Last Glacial Maximum. In calibrated form, this was about 22,000–23,000 yr ago.

14C CHRONOLOGY OF THE INTERPLENIGLACIAL Cultural and faunal remains of all kinds become far more scattered and scarce below Stratum 127 in a series of yellowish, light brown, colluvial, sandy, or clayey silts at the foot of the inner cave slope (Figure 11). However, flecks and chunks of charcoal are present, in association with occasional stone artifacts and bone fragments. From Stratum 128, a chunk of alder wood charcoal (identified by L Zapata) yielded a date of 27,580 ± 210 BP. Calibrated according to CALPAL 2001 by O Jöris (personal communication), this determination is equivalent to 29,332 ± 402 BC. This date corresponds radiometrically to early Gravettian levels (with Noailles burins) in the Basque cave sites of Antoliñako and Amalda, as well as to the culturally indistinct Level 7 at El Rascaño in the montane zone of the next river valley to the west of the Asón (Aguirre et al. 2000). It also overlaps with several early Aurignacian dates from Cueva Morín (Stuckenrath 1978), all of which have large standard deviations and could be “too young.” The El Mirón Stratum 128 date, associated with no diagnostic artifacts, simply proves that humans did, at least occasionally, visit the cave during terminal Aurig- nacian or Gravettian times, as also suggested by terminus ante quem TL dates of about 26,000 BP on calcite covering engravings in the nearby cave of Venta de la Perra (Moure and González 2000a). Lion remains have been tentatively identified by J Altuna in Stratum 128. Finally, the lowest level reached to date (Stratum 130) produced 2 flake denticulates, a few items of dÈbitage and bone fragments, and a 14C date on a chunk of conifer charcoal (possibly pine, according to anthracologist L Zapata) about 1 m below the Stratum 128 sample. The AMS determination of 41,280 ± 1120 BP is clearly within the terminal Mousterian time range, not only for Cantabria but also for Catalunya, the other region of Spain where a transition to the earliest Aurignacian is also proven to have taken place during about 40,000–38,000 BP uncalibrated (see Straus 1997, with references). Indeed, the MirÛn date is very similar (within the respective standard deviations) to the AMS dates for the terminal Mousterian (Level 20) at nearby El Castillo Cave (Cabrera et al. 1996) and at the Asturian rockshelter site of La Vi~~~~~~~~~~~~~na, Level XIII basal (J Fortea 2001), as well as at the Catalan sites of Romaní and L’Arbreda (Straus 1997). The CALPAL 2001 calibrated version of the Mirón Stratum 130 date is 41,485 ± 1062 BC, implying a 1000–3000 yr error at about the time of 56 L G Straus, M González Morales

Figure 11 Pre-Solutrean 14C dates for El Mirón Cave El Mirón Cave & the 14C Chronology of Cantabrian Spain 57

“the transition”, which is in line with the comparison of 14C and uranium-series dates from Romaní (Bischoff et al. 1994). Thus, El Mirón joins the growing list of Spanish sites that attest to the Mid- dle–Upper Paleolithic transition and it is hoped that one day, in another, more densely occupied part of the cave, richer (and perhaps even older) Mousterian occupation layers will be found.

CONCLUSION As it stands, the record of 14C dates from El Mirón Cave is one of the largest, most complete, and most systematically developed in Europe. Despite some major depositional hiati—first realized thanks to the dating program—the cultural-historical/chronostratigraphic sequence in this cave is remarkably complete due to the attractive and strategic nature of the cavity itself, its stability, and relative lack of internal erosion. The El Mirón sequence is quickly becoming one of the new reference sites for the Middle–Upper Paleolithic transition, for the late Upper Paleolithic, and for the post-Pleistocene prehistory of Cantabrian Spain. The participation of the Geochron labs is integral to the success of the El Mirón excavation and to that of other associated research projects in the RÌo Asón Basin. Use of a single, high-quality dating lab allows for strict comparability among levels and sites. The high standards of professionalism at Geochron we all owe to a great scientist, Hal Krue- ger, whose memory we salute.

ACKNOWLEDGEMENTS The El Mirón research has been principally funded by the US National Science Foundation and the Fundación M Botín of Santander, with other grants from the L S B Leakey Foundation, the National Geographic Society, Ministerio de Educación y Cultura, Gobierno de Cantabria, and University of New Mexico. Invaluable material support has been provided by the town and school of Ramales de la Victoria, the collective of former employees of the Sociedad de Trefilería, Universidad de Cant- abria, and Investigaciones Cibernéticas SA. Special thanks go to Alex Cherkinksy for most of the dates and to Olaf Jöris for the CALPAL calibrations. The maps and sections were redrafted by Ron Stauber from originals by Straus; Ariane Oberling Pinson drafted the charts. Finally, we wish to heartily thank our superb field and lab crews, collaborating scientists, and Joaquin (“Pencho”) Eguizába, irrepressible Covalanas Cave guide and friend of the El Mirón Project.

REFERENCES Aguirre M, López J, Sáenz de Buruaga A. 2000. Medio Bischoff J, Ludwig K, García J, Carbonell E, Vaquero M, ambiente, industrias y poblamiento prehistórico en Stafford T, Jull AJT. 1994. Dating of the basal Aurig- Urdaibai (Gemika, Bizkaia) del Würm reciente al Ho- nacian sandwich at Abric Romaní (Catalunya, Spain) loceno medio. Illunzar 4:13–38. by radiocarbon and uranium-series. Journal of Ar- Alvarez E, Jöris O. 1998. El significado cronológico de chaeological Science 21:541–51. algunas especies de fauna fría durante el Tardiglaciar Cabrera V, Bernaldo de Quirós F. 2000. Excavaciones ar- en la Península lbérica. Zephyrus 51:61–86. queológicas en la Cueva del Castillo, 1980–1999. In: Arias P, Altuna J, Armendariz A, González J, Ibáñez J, Ontañón R, editor. Actuaciones Arqueológicas en Ontañón R, Zapata L. 1999. Nuevas aportaciones al Cantabria, 1984–1999. Santander: Gobierno de Cant- conocimiento de las primeras sociedades productoras abria. p 23–34. en la región cantabrica. Saguntum-PLAV Extra 2:549– Cabrera V, Valladas H, Bernaldo de Quirós F, Hoyos M. 57. 1996. La transition Paléolithique moyen- Arias P, González C, Moure A, Ontañón R. 2000. Estudio Paléolithique supérieur à El Castillo: nouvelles data- integral del complejo arqueológico de La Garrna. In: tions par le carbone-14. Comptes-Rendus de I’Ac- Ontañón R, editor. Actuaciones Arqueológicas en adémie des Sciences de Paris Series IIa (322):1093–8. Cantabria, 1984–1999. Santander: Gobierno de Cant- Clark GA. 1976. El Asturiense Cantábrico. Madrid: Bib- abria. p 271–7. liotheca Praehistorica Hispana 13. Apellániz J. 1971. El Mesolítico de la Cueva de Tarrerón Crane H, Griffin J. 1960. University of Michigan radio- y su datación por el C-14. Munibe 23:91–104. carbon dates V. Radiocarbon 2:31–48. 58 L G Straus, M González Morales

Fortea J. 2001. El Paleolítico superior en Galicia y As- Serna MR. 2000. Excavaciones en la estación megalítica turias. In: Noiret P, editor. Paléolithique supérieur Eu- de Alto de Guriezo-Hayas. In: Ontañón R, editor. Ac- ropéen. Bilan Quinquennal 1996–2001. ERAUL 97. p tuaciones Arqueológicas en Cantabria, 1984–1999. 149–60. Santander: Gobierno de Cantabria. p 197–202. Fortea J. 2002. Trente-neuf dates C14-SMA pour l’art Straus LG. 1991. SW Europe at the Last Glacial Maxi- pariétal paléolithique des Asturies. Bulletin de la So- mum. Current Anthropology 32:189–99. ciété Préhistorique Ariége-Pyrénées 57:7–28. Straus LG. 1992. Iberia before the Iberians. Albuquer- Freeman LG. 1988. Dimensions of research at El Juyo. que: University of New Mexico Press. In: Dibble H, Montet-White A, editors. Upper Pleis- Straus LG. 1997. The Iberian situation between 40,000 tocene Prehistory of Western Eurasia. Philadelphia: and 30,000 BP in light of European models of migra- University of Pennsylvania Museum. p 1–39. tion and convergence. In: Clark GA, Willermet C, ed- Freeman LG. 1996. Obermaier y Altamira: las nuevas ex- itors. Conceptual Issues in Modern Human Origins cavaciones. In: Moure A, editor. El Hombre Fósil: 80 Research. New York: Aldine de Gruyter. p 235–52. Años Después. Santander: Universidad de Cantabria. Straus LG. 2001. Solutrean. In: Peregrine P, Ember M, p 249–69. editors. Encyclopedia of Prehistory, Volume 4. New García M. 2001. Comportamiento gráfico durante el Pa- York: Kluwer Academic/Plenum. p 328–50. leolitico superior en el Alto Asón: análisis de los dis- Straus LG, Clark GA. 1986. La Riera Cave. Tempe: An- positivos iconográficos rupestres. PhD dissertation. thropological Research Papers 36. Victoria: Department of Geography, Prehistory, and Straus LG, González Morales M. 2003. The Mesolithic in Archeology, Universidad del Pais Vasco, Victoria. the Cantabrian interior: fact or fantasy? In: Larsson L González Morales M. 1995. La transición al Holoceno en et al., editors. Mesolithic on the Move. Oxford: Oxbow la Región Cantábrica: en contrastste con el modelo del Books. p 359–68. Mediterráneo español. In: Villaverde V, editor. Los Ul- Straus LG, González Morales M, Farrand W, Hubbard W. timos Cazadores. Alicante: Instituto de Cultura Juan 2001. Sedimentological and stratigraphic observa- Gil-Albert. p 63–78. tions in El Mirón: a late Quaternary cave site in the González Morales M, Straus LG. 2000. La Cueva del Cantabrian Cordillera. Geoarchaeology 16:603–30. Mirón (Ramales de la Victoria, Cantabria): excava- Straus LG, González Morales M, García-Gelabert MP, ciones 1996–1999. Trabajos de Prehistoria 57:121– Fano M. 2002a. The late Quaternary human uses of a 33. natural territory: the case of the Rio Asón drainage International Study Group. 1982. An interlaboratory (eastern Cantabria Province, Spain). Journal of Ibe- comparison of radiocarbon measurements in tree rian Archaeology 4:21–61. rings. Nature 298:619–23. Straus LG, González Morales M, Fano M, García-Gelab- Mariezkurrena C. 1990. Dataciones absolutes para la ar- ert MP. 2002b. Last Glacial human settlement in east- queología vasca. Munibe 42:287–304. ern Cantabria. Journal of Archaeological Science 29: Mariezkurrena K, Altuna J. 1995. Fauna de mamíferos 1403–14. del yacimiento costero de Herriko Barra. Munibe 47: Stuckenrath R. 1978. Dataciones de carbono-14. In: 23–32. González Echegaray J, Freeman LG, editors. Vida y Moure A, González Sainz C. 2000a. Cronología del arte Muerte en Cueva Morín. Santander: Institución Cul- paleolitico cantábrico: últimas aportaciones y estado tural de Cantabria. p 215. actual de la cuestión. In: Oliveira Jorge V, editor. Ac- Stuiver M, Reimer P, Bard E, Beck J, Burr G, Hughen K, tas, III Congesso de Arqueología Peninsular. Volum e Kromer B, McCormac G, van der Plicht J, Spurk M. 2. Porto: ADECAP. p 461–73. 1998. INTCAL98 radiocarbon age calibration, Moure A, González Sainz C. 2000b. Documentación del 24000–0 cal BP. Radiocarbon 40:1041–83. arte rupestre en el sector central de la costa cantábrica. Yarritu MJ, Gorrochategui X. 1995. El megatitismo en el In: Ontañón R, editor. Actuaciones Arqueológicas en Cantábrico oriental. In: Primeros Agricultores y Ga- Cantabria, 1984–1999. Santander: Gobierno de Cant- naderos en el Cantábrico y Alto Duero. Cuadernos de abria. p 227–9. Sección: Prehistoria y Arqueología 6:155–98. Ruiz J, Smith P. 2001. El yacimiento del Cubio Redondo Zapata L. 1995. La excavación del depósito sepulcral cal- (Matienzo, Ruesga): una estación mesolítica de mon- colÌtico de la Cueva Pico Ramos. Munibe 47:35–90. taña en Cantabria. Munibe 53:31–55. RADIOCARBON, Vol 45, Nr 1, 2003, p 59–73 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

14C ABSOLUTE CHRONOLOGY OF PYRAMID III AND THE DYNASTIC MODEL AT PACHACAMAC, PERU

Adam MichczyÒski1,2 • Peter Eeckhout3 • Anna Pazdur1,4

ABSTRACT. Pachacamac, covering an area of about 600 hectares (ha) near the Pacific shore, is one of the largest and most important archaeological sites in Peru. Most of the monumental adobe-made buildings of the later pre-Inca period (or Late Intermediate Period, about 10th–15th century AD) are so-called pyramids with ramps (the role of the ramps has been interpreted in different ways). Precise dating of the pyramids appears as a crucial step in defining the functions of Pachacamac in pre-Inca times. In this paper, we present the results obtained from 3 field campaigns at Pyramid III, one of the biggest buildings of the site. A total of 24 radiocarbon datasets from 4 different laboratories will help us to place the various steps of development of Pyramid III on a timescale, defined on the basis of the excavations. More absolute dates are available from another pyramid with ramps, which allow us to make comparisons and propose a new model of interpretation for the Pachacamac site during the Late Intermediate Period (LIP).

INTRODUCTION Pachacamac is one of the biggest and most important sites of the ancient Andes. It is situated 30 km south of Lima and about half a km from the Pacific Ocean, on the right bank of the LurÌn river and close to its mouth (see Figure 1). Covering an estimated area of about 600 hectares (ha), its permanent occupation probably began during the first centuries of our era and continued until the Spanish Conquest and forced abandonment in AD 1535 (Eeckhout 1999a). Furthermore, Pachacamac is especially important for the history of Peruvian archaeology because it was there in 1896 that the German archaeologist Max ‹hle made the first scientific excavation on Peruvian soil and uncovered a stratigraphy that has formed the basis for all subsequent prehistoric chronology in the Central Andes (‹hle 1903; Menzel 1977). Four main successive cultural stages can be distinguished on the basis of excavated material: the Lima Period (about AD 200–550), the Wari Period (about AD 550– 900), the Ychsma Period (about AD 900–1470), and the Inca Period (about AD 1470–1533). Sur- prisingly, very few absolute dates are available from the site (Eeckhout 1999a; Shimada 1991; ZiÛ≥kowski et al. 1994) and some of them are without a precise documented provenience. This helps to explain why so little progress has been made in the chronology at Pachacamac since ‹hle’s pioneer work. Refinement of the chronology is one of the goals of the Ychsma Project5, especially for what concerns the post-Wari occupation at the site. Indeed, during the Ychsma Period (or Late Inter- mediate Period=LIP), the site experienced a remarkable growth seen in the construction of a number of pyramids with ramps. These buildings all have the same ground plan, which comprises one or more multilevel platforms linked by a ramp to a lower-level, rectangular, walled enclosure, with restricted access. A series of rooms is arranged in a “U” shape around the top of the platform. Adjoining these are other structures, usually interpreted as storerooms, living quarters, kitchen courtyards, etc. (Bueno Mendoza 1982; Eeckhout 1999a; Franco Jordan 1998; JimÈnez Borja 1985; Paredes Botoni 1988). There are 14 of these pyramid complexes, taking up about 75 ha (i.e., more

1Institute of Physics, Department of Radioisotopes, Silesian University of Technology, Krzywoustego 2, 44-100 Gliwice, Poland. 2Corresponding author. Email: [email protected]. 3FacultÈ de Philosophie and Lettres, UniversitÈ Libre de Bruxelles, Av. F. Roosevelt 50, 1050 Brussels, Belgium. Email: [email protected]. 4Email: [email protected]. 5The Ychsma Project (Archeological Investigation and Restoration Studies at Pachacamac) has been designed to answer questions about the function, development, and influence of Pachacamac during the Late Prehispanic Periods. Within the framework of a convention between the Instituto Nacional de Cultura del Per˙ and the UniversitÈ Libre de Bruxelles, the project team led by Peter Eeckhout has made excavations and surveyed the site since 1999.

59 60 A MichczyÒski et al.

B A

Figure 1 Location of Pachacamac in Peru (inset, top right) and location of Pyramid II (A-light grey area) and Pyramid III (B-dark grey area) within the site. Other elements represent main structures of the monumental area. than one-third of the monumental part of the archaeological area). Considering the importance of pyramids at the site, it is obvious that if we are to understand Pachacamac during the LIP, the reasons for the building of these pyramids must be established, who occupied them, whether they were occupied simultaneously, the duration of the occupation of each, and their relationships with one another and with the outside world. One current hypothesis proposes that the Pachacamac pyramids were outposts or embassies of different foreign ethnic groups affiliated with the cult of the site’s tutelar deity, and/or temples serving kin of the same deity. Pyramids in outlying areas could be local temples corresponding to each ethnic group represented at Pachacamac. This religious network would have had economical, and possibly political, implications (Agurto Calvo 1984; Bueno Mendoza 1982; Burger 1988; Franco 1998; Hyslop 1990; JimÈnez Borja 1962–63, 1985; JimÈnez Borja and Bueno Mendoza 1970; Keatinge 1988; Negro 1977; Paredes Botoni 1988, 1990a; Patterson 1983; Rostworowski 1972, 1989, 1992, 1993). Such a model is referred to here as the embassy model. On the basis of field research at Pachacamac and in the LurÌn Valley, an alternative interpretation has been proposed in which the pyramids were the palaces of local chiefs who succeeded one other according to the rules of their particular dynasty (Eeckhout 1999a, 1999b, 1999–2000, 2000a). This is the palace model. The building and the sequence of occupation are crucial in the sense that the embassy model is only acceptable if all pyramids were in use at the same time and particularly at the apogee of the Ychsma polity, just before the Inca conquest of the Central Coast (Bueno Mendoza 1974–75; Franco 1998; Hyslop 1990; JimÈnez Borja 1985; JimÈnez Borja and Bueno Mendoza 1970; Paredes Botoni 1988). This does not mean that all pyramids were actually built during a single Chronology of Pyramid III at Pachacamac, Peru 61 phase of the LIP but that most can be shown to be occupied simultaneously. In the palace model, the pyramids should be shown to have been constructed one after the other and to have functioned successively according to the dynastic interpretation. Precise dating of the pyramids, thus, appears to be a crucial step in defining of the functions of the whole site of Pachacamac in pre-Inca times. In this paper, we present the results obtained from 3 field campaigns at Pyramid III, one of the biggest buildings of the site. A total of 24 14C datasets from 4 different laboratories will help us to place the various steps of development of Pyramid III on a timescale, defined on the basis of the excavations. It has to be noted that more absolute dates are available, either from other (non-monumental) sectors of Pyramid III and from another pyramid with ramps.

Figure 2 Plan of Pyramid III complex (with indication of partial plan in Figure 4). Letters A, B, and C indicate 3 main pyramids of the compound. The lower part of the figure presents a three-dimensional reconstruction of the whole (after R Franco Jordan 1998).

PYRAMID III AT PACHACAMAC Pyramid III is located in the northeast sector of the site (see Figures 1 and 2). On the east and north, it is bound by large, open spaces; on the west, by rooms associated with Pyramid II; and on the south, by the end of the East street (the main part of the street being displaced by the pyramid). It dominates the whole site (with the exception of the Temple of the Sun) and covers a surface area of about 16,000 m2. The building is composed of 2 main pyramidal complexes (A and B), an adjoining structure to the northwest (C), and 2 large plazas (II and III) surrounded by walls. The south portion of the pyramid (complexes A and B) is built on a promontory called “hill Z”; it dominates the 62 A MichczyÒski et al. remaining structures by between 2–7 m, according to their level. We shall here focus on the southern part of this complex. Results of 3 excavation campaigns (led in 1993, 1995, and 1999) indicate that these buildings were occupied by restricted elites who organized feasts and other punctual meetings, thus, implying a huge number of participants (Eeckhout 1999b, 2001). Pyramids are also sites of production (of textiles, ceramics, etc.), breeding (of guinea pigs), and storage (of agricultural and possibly other products). Previous research has put in evidence 3 successive phases within the monumental compound (see Figure 3). The last two are pyramids with ramps that have been constructed successively (first, Pyramid III-B and then, Pyramid III-A). Traces of ritual abandonment including the covering of certain structures with selected sand, the offering of deposits, and the sealing of access points have been observed in Pyramids A and B, as well as in Phase D (the oldest one). This evidence, jointly with the available 14C dates from the first 2 campaigns, have led us to suggest that Pyramids B and A had been constructed, occupied, and abandoned one after the other. The 1999 campaign strategy has been designed to check this hypothesis through further excavations and rec- ollection of more 14C samples from the most crucial zones (Figure 4). Within the framework of this paper, we develop an in-depth analysis of whole samples that should allow us to propose an absolute chronology to the archaeological sequence at Pyramid III.

Figure 3 Sketch of the successive constructive D, B, and A phases of Pyramid III. Phases A and B correspond with parts of Pyramid III indicated in Figure 2 by letters A and B. It can be seen that Phase D (the earliest one) has been covered by Phase B, which in turn has been partially covered by Phase A (the latest one). In absolute terms, occupation of Phase B corresponds with STEP 2 and occupation of Phase A corresponds with STEP 4.

METHODS AND RESULTS OF 14C DATING 14C dating was carried out using organic material (wood, charcoal, grains, plant remains, etc.). Most of the samples were dated in the Gliwice Radiocarbon Laboratory using the gas proportional counting method (Lab. No. Gd-; Pazdur et al. 2000) with the standard sample pretreatment (Pazdur and Pazdur 1986). Only 1 sample (GdA-90; Table 1) was dated with the AMS technique (Goslar and Czernik 2000). All conventional 14C dates are corrected for δ13C according to the Stuiver and Polach procedure (1977). Complete information about the samples and the results from 14C dating appear in Table 1. Apart from the Gliwice Radiocarbon Laboratory dates, there are 4 dates from the Utrecht Chronology of Pyramid III at Pachacamac, Peru 63

Figure 4 Partial plan of Pyramid IIIA–B with sample provenience locations for 14C dating van de Graaff Laboratory, the Netherlands (Lab. Code UtC-) and 1 date from the Royal Institute of Cultural Heritage, Belgium (Lab. Code IRPA-). The 14C dates have been calibrated using the program GdCALIB, developed in the Gliwice Radiocarbon Laboratory (Pazdur and MichczyÒska 1989) with a new calibration curve (Stuiver et al. 1998). The calendar age is represented by 68% and 95% narrowest confidence intervals.

PROPOSED MODEL OF THE CHRONOLOGY Excavations have allowed us to define 7 successive steps on the basis of the following: both architectural analyses, stylistic features of encountered material, and absolute dating. Here are the 7 successive steps: • STEP 1: Refuse deposit on Hill Z, with association of various kinds (architecture, domestic occupation) • STEP 2: Construction and occupation of Pyramid B or Phase B (about 30 yr maximum) • STEP 3: Voluntary abandonment of Pyramid B (a very short event) • STEP 4: Construction and occupation of Pyramid A or Phase A (about 30 yr maximum) • STEP 5: Voluntary abandonment of Pyramid A (very short event, before Inca conquest, i.e., AD 1470) • STEP 6: Intrusive reoccupation of funerary (between AD 1470–1533?) and domestic nature (after AD 1533, i.e., date of Spanish conquest?) • STEP 7: Definitive abandonment and looting (after AD 1533?) 64 A MichczyÒski et al. 95% narrowest confidence intervals [686 AD, 1302 AD] 94.77% [1027 AD, 1401 AD] 95.47% [1265 AD, 1414 AD] 95.45% [1387 AD, 1471 AD] 76.80% [1306 AD, 1354 AD] 18.32% [1340 AD, 1397 AD] 57.72% [1291 AD, 1331 AD] 37.52% .48% [1405 AD, 1445 AD] 95.40% Pachacamac, Peru. STEP 3/4 indicates the border between 68% narrowest confidence intervals [892 AD, 1223 AD] 67.13% [1158 AD, 1305 AD] 53.67% [1158 AD] 3.10% AD, 1137 [1124 [1067 AD, 1083 AD] 3.77% [1291 AD, 1326 AD] 29.28% [1333 AD, 1338 AD] 5.15% [1299 AD, 1319 AD] 24.02% C age [BP] Calibrated age [BC/AD] 14 Gd-16039120± 770 [1355 AD, 1386 AD] 7.56% Gd-17040180± 980 [1231 AD, 1238 AD] 1.20% Gd-15132 650 ± 65 [1347 AD, 1392 AD] 38.75% Gd-12263 510 ± 50 [1400 AD, 1441 AD] 63.21% Gd-11548 490 ± 30 [1418 AD, 1438 AD] 69 IRPA-1131 640 ± 35 [1352 AD, 1388 AD] 44.19% W W W W W W ′ ′ ′ ′ ′ ′ wood lucuma C and calibrated ages of the samples from Pyramid III, 14 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 ′ ′ ′ ′ ′ ′ C = –20.77‰ C = –23.38‰ C = –25.69‰ C = –26.11‰ C = –18.61‰ 13 13 13 13 13 Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 Material: charcoal δ Lat 12°12 Material: charcoal and grains δ Material: piece of Material: grains-carbonized corn δ Material: charcoal δ Material: charcoal δ Sample Lab Nr STEP 1 1 III-12-j-9 PAC 2 III-12-j-6 PAC 5 III-10-4 PAC 7 III-22-a-2 PAC 4 III-6-u-9 PAC STEP 2 6 III-22-a-3 PAC STEP 3 and 4 Table 1 Table Conventional Chronology of Pyramid III at Pachacamac, Peru 65 95% narrowest confidence intervals [1423 AD, 1518 AD] 88.73% [1382 AD, 1440 AD] 53.39% [1303 AD, 1369 AD] 42.05% [1575 AD, 1626 AD] 15.54% [1431 AD, 1521 AD] 80.08% [1556 AD, 1631 AD] 7.30% [1277 AD, 1525 AD] 88.12% [1910 AD, 1950 AD] 6.60% [1830 AD, 1890 AD] 4.40% [1720 AD, 1820 AD] 22.10% [1450 AD, 1710 AD] 62.40% [1381 AD, 1439 AD] 50.60% [1303 AD, 1369 AD] 44.86% [1286 AD, 1453 AD] 95.19% 68.71% [1596 AD, 1620 AD] 6.90% acamac, Peru. STEP 3/4 indicates the border between 68% narrowest confidence intervals [1326 AD, 1348 AD] 22.54% [1439 AD, 1489 AD] 65.58% [1301 AD, 1373 AD] 33.77% [1730 AD, 1810 AD] 17.00% [1610 AD, 1680 AD] 22.50% [1510 AD, 1600 AD] 25.40% [1390 AD, 1430 AD] 42.99% [1324 AD, 1349 AD] 25.28% [1387 AD, 1427 AD] 30.58% [1306 AD, 1354 AD] 34.59% C age 550 ± 50 560 ± 75 14 [BP] Calibrated age [BC/AD] Gd-11554 545 ± 50 [1392 AD, 1432 AD] 45.48% Gd-12274 ± 35 425 [1438 AD, 1479 AD] Gd-12270 ± 35 410 [1604 AD, 1607 AD] 2.56% Gd-15136105± 540 [1378 AD, 1445 AD] 34.44% Gd-15134 ± 75 255 [1930 AD, 1950 AD] 3.20% Gd-12261 Gd-15150 W W W W W W ′ ′ ′ ′ ′ ′ C and calibrated ages of the samples from Pyramid III, Pach 14 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 ′ ′ ′ ′ ′ ′ (Continued) C = –26.15‰ C = –29.07‰ C = –27.67‰ C = assumed to be equal –28‰ C = –18.98‰ C = –29.90‰ 13 13 13 13 13 13 Material: wood δ Lat 12°12 Material: charcoal δ Lat 12°12 Material: wood δ Material: charcoal and plant remains δ Material: charcoal δ Material: plant remains δ Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 10 III-24-a/b-4 PAC 8 III-12-a-á-5 PAC Sample Lab Nr 9 III-30-a-2 PAC 11 III-25-m-2a PAC 12 III-25-f-3 PAC 13 III-23-Hornacina 2 PAC 3 and STEP 4 Table 1 Table Conventional 66 A MichczyÒski et al. 95% narrowest confidence intervals [1387 AD, 1434 AD] 56.92% [1306 AD, 1354 AD] 38.26% AD] 28.31% [1381 AD, 1411 [1302 AD, 1370 AD] 67.01% [1391 AD, 1436 AD] 72.90% [1325 AD, 1349 AD] 22.53% [1298 AD, 1405 AD] 95.52% [1331 AD, 1341 AD] 5.69% 67.81% [1397 AD, 1443 AD] 89.59% 68.33% [1420 AD, 1478 AD] 95.45% 23 AD] 57.54% 96 AD] 13.07% Pachacamac, Peru. STEP 3/4 indicates the border between 68% narrowest confidence intervals [1329 AD, 1343 AD] 23.36% [1314 AD, 1353 AD] 47.55% [1333 AD, 1338 AD] 10.11% [1342 AD, 1368 AD] 27.17% [1304 AD, 1330 AD] 27.74% C age 14 [BP] Calibrated age [BC/AD] Utc-4463 552 ± 33 [1395 AD, 1420 AD] 44.71% Utc-4467 586 ± 32 [1388 AD, 1404 AD] 20.58% Gd-12269 540 ± 30 [1399 AD, 14 Gd-12262 610 ± 40 [1384 AD, 13 Gd-12271 510 ± 35 [1409 AD, 1434 AD] Gd-11547 450 ± 30 1428 AD, 1455 AD] W W W W W W ′ ′ ′ ′ ′ ′ C and calibrated ages of the samples from Pyramid III, (Continued) 14 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 ′ ′ ′ ′ ′ ′ C = –26.2‰ C = –25.69‰ C = –28.55‰ C = –27.46‰ C = –28.26‰ C = –27.05‰ 13 13 13 13 13 13 Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 Lat 12°12 Material: charred seeds δ Material: charred corn ear δ Material: wood δ Material: charcoal δ Material: wood δ Material: plant remains δ Sample Lab Nr 14 III-3-15 PAC 15 III-4-c20-7 PAC 16 III-30-c-1 PAC STEP 3/4 17 III-6-u-4 PAC STEP 4 18 III-14-f-2c PAC 19 III-12-h-2 PAC STEP 3 and 4 Table 1 Table Conventional Chronology of Pyramid III at Pachacamac, Peru 67 95% narrowest confidence intervals [1558 AD, 1630 AD] 35.02% [1444 AD, 1525 AD] 60.14% [1406 AD, 1516 AD] 91.57% [1477 AD, 1637 AD] 95.29% [1555 AD, 1631 AD] 27.45% [1434 AD, 1526 AD] 68.17% 3/4 indicates the border between STEP [1412 AD, 1443 AD] 95.19% 5% [1422 AD, 1480 AD] 95.12% 68.67% [1599 AD, 1616 AD] 3.74% 8 AD] 15.89% 1617 AD] 15.44% 68% narrowest confidence intervals [1453 AD, 1516 AD] 52.96% [1564 AD, 1604 AD] 28.01% [1489 AD, 1523 AD] 24.66% [1500 AD, 1513 AD] 7.78% [1442 AD, 1494 AD] 50.99% from Pyramid III, Pachacamac, Peru. STEP C age 485 ± 25 [1422 AD, 1438 AD] 68.70% 14 [BP] Calibrated age [BC/AD] Utc-3686 379 ± 31 [1599 AD, GdA-90 (KIA13736) Gd-11552 445 ± 30 [1432 AD, 1461 AD] 68.7 Gd-12273 345 ± 30162 AD, [1606 Gd-12268 400 ± 40 [1600 AD, 1614 AD] 9.71% Utc-4464 450 ± 43 [1424 AD, 1472 AD] W W W W W W ′ ′ ′ ′ ′ ′ C and calibrated ages of the samples 14 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 S Long 76°54 ′ ′ ′ ′ ′ ′ (Continued) C = –28.14‰ C = –27.40‰ C = –30.02‰ C = –11.1‰ 13 13 13 13 Material: maize grains Lat 12°12 Material: plant remains Lat 12°12 Material: charcoal δ Lat 12°12 Material: charcoal δ Lat 12°12 Material: charcoal δ Material: charred seeds δ Lat 12°12 Lat 12°12 25 III-10-Cx1 PAC STEP 6 24 III-12-a-2(H6) PAC Sample Lab Nr STEP 5 20 III-24-k-2C PAC 22 III-24-h-2 PAC 21 III-24-h-2B PAC 23 III-6-g-5-H21 PAC 3 and STEP 4 Table 1 Table Conventional 68 A MichczyÒski et al.

As can be seen, Phase B and Phase A are very short and one succeeds the other directly. It has to be added that Phase A partially covers Phase B. The hypothesis is that Pyramids B and A are successive palaces of kings that succeeded one another following a dynastic-like rule. Hereafter (see Table 1 and Figure 5), we propose an ideal relationship between the results of 14C dating and the successive steps.

STEP 1 Gd-16039 Gd-17040 Gd-15132 IRPA-1131 Gd-12263 STEP 2 Gd-11548 Gd-11554 Gd-12274 Gd-12270 Gd-15136 Gd-15134 Gd-12261 Gd-15150 Utc-4463 Utc-4467 Gd-12269 Gd-12262 STEP 3/4 Gd-12271 STEP 4 Gd-11547 Gd-11552 STEP 5 Gd-12273 Gd-12268 Utc-4464 GdA-90 STEP 6 Utc-3686

600 800 1000 1200 1400 1600 1800 2000 Calendar year AD Figure 5 Successive steps of the proposed model of the Pyramid III chronology with calendar (calibrated) ages of individual samples belonging to these steps. STEP 3/4 indicates the border between STEP 3 and STEP 4.

STATISTICAL ANALYSIS OF RESULTS Figure 6 shows probability distributions obtained from 14C dates presented in Table 1 with division of these on the steps according to proposed model. We decided to exclude from our analysis Sample PAC III-25-f-3 (date Gd-15134), which we assume to be aberrant. All distributions, except STEP 5, are cumulative (composite) probability distributions and they were calculated by summarizing on probability distribution of calendar age of samples, which belongs to the same phase (STEP). Prob- ability distribution for STEP 5 was calculated using another method. Based on what we know from Chronology of Pyramid III at Pachacamac, Peru 69 the ethno-historical information, this step—voluntary abandonment of Pyramid A—must have been a very short event (<1 yr). Therefore, we assumed that the 14C dates from STEP 5 date the same event and we calculated the probability distribution of the weighted mean of these dates. Results of the 14C dating of samples for Pyramid II are presented in the Table 2.

Figure 6 Probability distributions of calendar ages of all samples, which belong to the same steps (phases).

Table 2 Results of 14C dating of samples for Pyramid II at Pachacamac (Paredes and Franco 1987). Samples dated by the Pontifica Universidad Catolica del Peru, Lima. Event 14C Age (BP) Context and material last Ychsma occupation level 600 ± 70 Upper platform of Pyramid II-A, square C4, wood lintel Ychsma occupation level 654 ± 80 Upper platform of Pyramid II-B, square D4, post buried in ground not precise 660 ± 160 Annex buildings, square E6, charred wood 70 A MichczyÒski et al.

We notice that the cumulative probability distribution for samples from Pyramid II show that this pyramid is older than Pyramid III. This result enables us to assume that Pyramid II and III functioned successively. The cumulative probability distribution for STEP 1 agrees with the proposed model of relationships and covers similar time intervals as the distribution for Pyramid II. STEP 1 represents a phase when hill Z was covered with refuse and free of any architecture; material there apparently accumulated during almost a century and was then used as constructive fill for Pyramid III-B at the beginning of 15th century. Because of wiggles on the calibration curve, the distribution for STEP 2 (Phase B) has two maxims of likelihood. However, we can assume that the limits of Phase B are defined by a second maximum of probability distribution (AD 1380–1460) and that the first maximum (AD 1320–1360) is only an effect of the wiggles. The border between STEP 1 and STEP 2, based on the probability distribution for STEP 1, agrees well with this based on the second peak of the probability distribution for STEP 2 and is equal to about AD 1400. On the other hand, probability distributions obtained for STEP 2 and STEP 4 show that Phases A and B are contempo- raneous or indistinguishable, but Phase B seems to start somewhat earlier than Phase A. The distribution for STEP 5 may be treated as the terminus ante quem age of Phase A (STEP 4) and clearly shows that this phase ends about AD 1460–1470. The same value arises from time intervals calculated on the basis of the probability distribution of STEP 4. It agrees very well with the supposed length of occupation and the idea of pre-Inca abandonment.

DISCUSSION Probability distribution of calibrated 14C dates seem to confirm the proposed chronology model of the 7 successive steps of building of Pyramid III at Pachacamac, formulated on the basis of both the architectural analysis and the stylistic features of encountered material. Even if there remain some difficulties in distinguishing Phases A and B, these could be explained by visits to the ancestors buried in Phase B by those dwelling in Phase A. Ethnohistoric accounts of such practices inform us that important deceased leaders, such as lords and emperors, were consulted on a regular basis through divinination practices and that food and vestments were offered—some probably burned—at these occasions (Cieza de LeÛn 1994; Cobo 1956; Valera 1968). Considering this context, one can imag- ine that even if the old palace was replaced by a new one and did not function anymore as the living rulers’ apartments, it was not totally abandoned. Architectural features seem to support such an interpretation since a narrow corridor between the elite residential rooms of Pyramid A and the main plaza of Pyramid B were discovered in 1999 (Eeckhout 2001). This corridor was built during Phase A and ritually sealed (with caches of offerings) at the end of that phase. Another point to be addressed is the surprisingly remote dating of Step 6, specifically Sample 23 (see Table 1). First, both Samples 23 and 24 come from intrusive burials containing local and Inca- influenced material, something that has led us to suggest that they were posterior to the abandonment of the pyramid and culturally related to the Late Horizon or Inca period of occupation of the site (Eeckhout 1999b). Second, Sample 23 is composed of plant remains found inside an in situ intact pot related to the Chimu-Inka style (Eeckhout 2000b: Figure 10). This means that from an archaeological point of view, the context of the sample is absolutely safe and secure. The only possibility then would be that the plant remains that were used for dating were older than the rest of the burial context, something that seems difficult to demonstrate since we have no more samples of the same kind. Nevertheless, there were other organic remains in the same context, so the future dating of these samples could help us solve the problem. Chronology of Pyramid III at Pachacamac, Peru 71

CONCLUSIONS AND FURTHER HYPOTHESIS Comparison of the dating of Pyramids II and III confirms the hypothesis of a successive occupation and abandonment of these buildings. This constitutes a total breakthrough in the current understanding of the function of the site of Pachacamac during the pre-Inca period. The length of the occupation of these buildings strengthens the hypothesis of a dynastic type of succession, with each king/ curaca having a new palace built upon the death of his predecessor. This hypothesis has the advantage of relating the archaeological field data to written sources from the 16th and 17th centuries, as they refer to how power is exercised in the Andes. It places Pachacamac in a specifically Andean setting [cf., the successive places of the Inca emperors in Cuzco (Rowe 1967) and the citadels of the different Chimu sovereigns in Chan Chan (Conrad 1981; Kolata 1982)]. It also agrees with the model of progressive secularization of Andean authority structures, which has been put forward recently by an increasing number of investigators (McEwan 1990; Moseley 1985; von Hagen and Craig 1998). If one pushes the generational hypothesis a little further, it also helps to explain the special distribution of the pyramids with ramps at Pachacamac. In fact, in the course of the architectural analysis made on the field (Eeckhout 1999a), it has become apparent that numerous pyramidal complexes were really composed of several pyramids. The study of the circulation system showed that at the heart of a single complex, there was never more than 1 main entrance (Eeckhout 2000a). In other words, only 1 of the pyramids was directly accessible from the exterior. Access to the other pyramids was subordinated to that of the main pyramid (cf., Pyramids I, VII) or else blocked by the construction of the latter (cf., Pyramids III, XII). On the basis of the results of the excavations in Pyra- mid III and the chronological evidence for Pyramids II and III, we would suggest that several pyramids in the same pyramidal complex form successive stages in construction, occupation, and voluntary abandonment, and correspond to the successive reigns of several kings. The size differences among the pyramids would correspond to the fluctuations of power across the years, according to whether the king had access to a larger or smaller work force. Pyramids that are apparently unfinished (IV, V, and VIII) would reflect the unexpected death of a king during the construction of the same, which would have then been interrupted. If one accepts this working hypothesis, it must be understood that there are 18 pyramids with central ramps at Pachacamac (some isolated, some included in multi-pyramidal compounds) and it seems clear that they were not all simultaneously functioning entities. Hence, 2 possible interpretations can be advanced: single generational succession or multiple generational succession. In single generational succession, it must be understood that 1 pyramid at a time was in use at Pachacamac and that the pyramids have been built, occupied, and abandoned in succession by kings who ruled alone over the site. The period covered by the total of the successive pyramids with ramps would be in the order of about 400–450 yr (15 completed pyramids multiplied by 25–30 yr). If, as all authors seem to agree, the construction of pyramids with ramps was interrupted by the Inca conquest (around AD 1470), this would place the first potential pyramid in the 11th century (i.e,. at the beginning of the LIP). On the other hand, multiple generational succession assumes that several lin- eages built their own pyramids in succession at Pachacamac and that certain kings were then con- temporaneous with one another. It is impossible in such a case, given the current state of research, to estimate the duration of the period covered by pyramids with ramps. It is, however, probable that the start of the building phases commenced later than that of the model for single generational succession. 72 A MichczyÒski et al.

These crucial issues are (in the present state of research) largely speculative but we are confident in the fact that in the future years the new data collected on the field at Pachacamac will shed new light on the prehistory of this important site of the Central Andean area.

ACKNOWLEDGEMENTS The statistical interpretation of the 14C results was financially supported by the Polish State Com- mittee for Scientific Research within research project No. 6 P04E 007 17. Fieldwork and 14C dating costs were sponsored by the Fonds de la Recherche Fondamentale Collective (Belgium), the Fonds National de la Recherche Scientifique (Belgium), The FacultÈ de Philosophie and Lettres de l’Uni- versitÈ Libre de Bruxelles, and a grant from the National Geographic Society Committee for Research and Exploration (Washington). We acknowledge the Instituto Nacional de Cultura del Per˙ for giving us the research permit at Pachacamac and all the excavation staff that worked with us on the field.

REFERENCES Agurto Calvo S. 1984. Lima Prehispanica. Lima: Munic- Arts of Africa, Oceania and the Americas. Norwich: ipalidad de Lima Metropolitana. Recent Research on the Americas, University of East Bueno Mendoza A. 1974–75. Cajamarquilla y Pachaca- Anglia. mac: dos ciudades de la costa central del Perú. Boletín Eeckhout P. 2000b. Investigaciones arqueológicas en la Bibliográfico de Antropología Americana Volum e Pirámide con rampa n° III de Pachacamac, costa cen- XXXVII(46):171–211. tral del Perú. Estudios Latinoamericanos 20:19–40. Bueno Mendoza A. 1982. El antiguo valle de Pachaca- Eeckhout P. 2001. La temporada 1999 de excavaciones mac: espacio, tiempo y cultura. Lima: Editorial de Los en Pachacamac. Paper presented at The 66th Annual Pinos. Meeting of the Society for American Archaeology, Burger RL. 1988. Unity and heterogeneity within the New Orleans. Chavín Horizon. In: Keatinge R, editor. Peruvian Pre- Franco Jordan R. 1998. La Pirámide con Rampa n°2 de history. Cambridge: Cambridge University Press. Pachacamac. Excavaciones y Nuevas Interpreta- p 99–144. ciones. Trujillo: Universidad Nacional de Trujillo, Cieza de León P. 1994. [1551] Crónica del Perú, primera Peru. parte. Lima: Pontifica Universidad Católica del Perú- Goslar T, Czernik J. 2000. Sample preparation in the Gli- Academia Nacional de la Historia. wice radiocarbon laboratory for AMS 14C dating of Cobo FB. 1956. [1653] Historia del Nuevo Mundo. Bib- sediments. Geochronometria 18:1–8. lioteca de Autores Espanoles. Madrid: Ediciones At- Hyslop J. 1990. Inca settlement planning. Austin: Uni- las. p 41–42. versity of Texas Press. Conrad GW. 1981. Cultural materialism, split inherit- Jiménez Borja A. 1962–63. El conjunto arqueolÛgico de ance, and the expansion of ancient Peruvian empires. Pachacámac. In: Jiménez Borja A, editor. Informe so- American Antiquity 46:3–26. bre los Sitios Arqueológicos de Lima. Lima: Junta De- Eeckhout P. 1999a. Pachacamac durant l’Intermédiaire liberante: 27–32. récent. Etude d’un site monumental préhispanique de Jiménez Borja A, Bueno Mendoza A. 1970. Breves notas la CÙte centrale du PÈrou. British Archaeological Re- acerca de Pachac·mac. Arqueología y Sociedad 4. ports. International Series 747. Oxford: Hadrian Lima: Museo de Arqueología y Etnología de la Uni- Books Ltd. versidad Nacional Mayor de San Marcos. p 13–25. Eeckhout P. 1999b. Pirámide con rampa n° III, Pachacá- Jiménez Borja A. 1985. Pachacámac. Boletín de Lima 38: mac. Nuevos datos, nuevas perspectivas. Boletín del 40–54. Instituto Frances de Estudios Andinos 28(2):169–214. Keatinge R. 1988. A summary view of Peruvian prehis- Eeckhout P. 1999–2000. The palaces of the lords of Ych- tory. In: Keatinge R, editor. Peruvian Prehistory. sma. An archaeological reappraisal of the function of Cambridge: Cambridge University Press. p 303–16. pyramids with ramps at Pachacamac, central coast of Kolata AL. 1982. Chronology and settlement growth at Peru. Journal of American Archaeology 17-18-19: Chan Chan. In: Moseley M, Kent CD, editors. Chan 217–54. Chan: Andean Desert City. Albuquerque: A School of Eeckhout P. 2000a. Pachacamac and the pyramids with American Research Book, University of New Mexico ramps. Prehispanic monuments and models on the Press. p 67–85. central coast of Peru. Paper presented at The Fifth McEwan GF. 1990. Some formal correspondances be- Symposium of the Saintsbury Research Unit for the tween the imperial architecture of the Wari and Chimu Chronology of Pyramid III at Pachacamac, Peru 73

cultures and their implications for the origin of the ar- dates. Radiocarbon 31(3):824–32. chitecture of Chan Chan. Latin American Antiquity Rostworowski de Díez Canseco M. 1972. Breve informe 1(2):97–116. sobre el Señorio de Ychma o Ychima. ArqueologÌa Menzel D. 1977. The archaeology of ancient Peru and PUC 13. Lima: Instituto Riva-Ag¸ero, Pontificia Uni- the work of Max Ühle. Berkeley: R.H. Lowie Museum versidad CatÛlica del Perú. p 37–51. of Anthropology, University of California-Berkeley. Rostworowski de DÌez Canseco M. 1989. Costa Peruana Moseley ME. 1985. The exploration and explanation of prehispánica. Lima: Instituto de Estudios Peruanos. early monumental architecture in the Andes. In: Don- Rostworowski de DÌez Canseco M. 1992. Pachacámac y nan CB, editor. Early Ceremonial Architecture in the el Señor de los Milagros. Una Trayectoria Milenaria. Andes. Dumbarton Oaks Research Library and Col- Lima: Instituto de Estudios Peruanos. lection. p 29–58. Rostworowski de DÌez Canseco M. 1993. Orígen reli- Negro S. 1977. Patrones de asentamiento Prehisp·nico gioso de los dibujos y rayas de Nasca. Journal de la en el Valle de Lurín. Unpublished Masters thesis. SociÈtÈ des AmÈricanistes 79: 189–202. Lima: Universidad Ricardo Palma. Rowe JH. 1967. What kind of settlement was Inca Paredes Botoni P. 1988. Pachacámac—pirámide con Cuzco? Ñawpa Pacha 5:59–76. rampa n°2. Boletín de Lima 55:41–58. Shimada I. 1991. [1903] Pachacamac archaeology. Ret- Paredes Botoni P. 1990. Pachacamac. In: Purin S, editor. rospect and prospect. Philadelphia: University of Inca-Perú. 3000 ans d’Histoire. Catalogue de l’expo- Pennsylvania Press. sition prÈsentÈe aux MusÈes Royaux d’Art et d’His- Stuiver M, Polach HA. 1977. Reporting of 14C data. Ra- toire de Bruxelles. Gent: Imschoot uitgevers. p 178– diocarbon 19(3):355–63. 95. Stuiver M, Reimer PJ, Braziunas TF. 1998. High-preci- Patterson TC. 1983. Pachacamac—an Andean oracle un- sion radiocarbon age calibration for terrestrial and ma- der Inca rule. In: Kvietok P D, Sandweiss D H, editors. rine samples. Radiocarbon 40(3):1127–51. Recent Studies in Andean Prehistory and Protohistory. ‹hle M. 1903. Pachacamac. Report of The William Pep- Papers from the Second Annual Northeast Conference per, M.D., LL.D. Peruvian Expedition of 1896. Phila- on Andean Archaeology and Ethnhohistory. Ithaca: delphia: Department of Archaeology of the University Latin American Studies Program, Cornell University. of Pennsylvania. p 159–76. Valera B. 1968. [1585–1589] Relación de las costumbres Pazdur A, MichczyÒski A, Pawlyta J, Spahiu P. 2000. antiguas de los naturales del Perú. In: Barba EF, editor. Comparison of the radiocarbon dating methods used Crónicas Peruanas de Interes Indigena Biblioteca de in the Gliwice Radiocarbon Laboratory. Geochro- Autores Españoles T. 209. Madrid: Ediciones Atlas: nometria 18:9–14. 135–203. Pazdur A, Pazdur MF. 1986. Aparatura pomiarowa Lab- Von Hagen A, Craig M. 1998. The cities of the ancient oratorium C-14 w Gliwicach. Doúwiadczenia kon- Andes. London and New York: Thames and Hudson. strukcyjne i eksploatacyjne (The measuring equip- Zió≥kowski M., Pazdur MF, Krzanowski A, MichczyÒski ment of the Gliwice Radiocarbon Laboratory. A. 1994. Andes. Radiocarbon database for Bolivia, Experience gathered in the construction and exploata- Ecuador and Peru. Warszawa-Gliwice: Andean Ar- tion). Zeszyty Naukowe Politechniki ålπskiej, Seria chaeological Mission of the Institute of Archaeology, Matematyka-Fizyka, Z. 46. Geochronometria 1:55– Warsawa University, and Gliwice Radiocarbon Lab- 69. oratory of the Institute of Physics, Silesian Technical Pazdur MF, MichczyÒska DJ. 1989. Improvement of the University. procedure for probabilistic calibration of radiocarbon RADIOCARBON, Vol 45, Nr 1, 2003, p 75–80 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

HIGH-PRECISION AMS 14C RESULTS ON TIRI/FIRI TURBIDITE

Thomas P Guilderson1,2 • John R Southon1,3 • Thomas A Brown1

ABSTRACT. Unleached aliquots of TIRI/FIRI turbidite were analyzed by accelerator mass spectronomy (AMS) over a timespan of 18 months. Individual analyses ranged from 18,090–18,245 yr BP with reported errors between 30–50 yr. The weighted average fraction modern (FM) of these 28 measurements is 0.10378 ± 0.00008 (which equates to 18,199 ± 8 yr BP) and the measurements show a 1 standard deviation scatter of 0.00044 (±35 yr). The fractional error of these results indicates reproducibility of individual measurements at the 4‰ (1σ) level, which is consistent with the quoted counting-statistics-based errors. Laboratories engaged in the determination of 14C results at reasonably high precision should consider taking advantage of the TIRI and FIRI sample materials in the role of process standards. Additional suites of high-precision data are necessary to refine the accuracy of these sample materials.

INTRODUCTION The desire to predict future climate change is forcing us to look back into the past at increasingly finer temporal resolution to determine the regional leads and lags of the climate system as it responds to a myriad of external forces at decadal to centennial frequencies. The underlying goal of these studies is to develop a better understanding of the forces and responses that result in natural climate changes (e.g., sub-Milankovitch solar forcing, internal oscillations, and cryosphere processes). The determination of such leads and lags requires both more detailed and more robust chronological control than has routinely existed for paleoclimate studies and a better understanding of the inherent and implicit errors in calibrating ages from a variety of materials and locations to a common temporal timescale. The Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratory has contributed not only to the development of individual paleoclimate studies and the interaction of climate and human civilization (e.g., Weiss et al. 2002) but also to the definition of the radiocarbon calibration timescale (e.g., Hughen et al. 1998, 2000). As a consequence of our high-resolution coral- radiocarbon work and in concert with enhancements to, and advances in the understanding of, our ion source (Southon and Roberts 2000; Brown et al. 2000), we have modified our data acquisition scheme and are now building up representative datasets that validate reasonably high-precision AMS-14C that for older (late Pleistocene) samples is comparable to the capabilities available from high-precision liquid scintillation counting (e.g., McCormac et al. 1998). The TIRI/FIRI turbidite results presented here are part of an ongoing, high-precision, coral-based Th/U 14C calibration project (e.g., Fairbanks et al. 2001) and were analyzed as unknowns concurrently with the coral samples to monitor/confirm the reported precision of the coral results.

METHODS An appropriate amount to yield 1.0 mg-carbon targets (~20 mg) of TIRI/FIRI turbidite was placed in individual vacutainers and evacuated to ≤1 × 10–3 Torr with gentle heating. A 0.5 ml aliquot of 85% phosphoric acid was injected into the vacutainer, after which the vacutainer was placed on a heating block at 90 °C for 1 hr. The resulting CO2 was cryogenically purified to remove water and transferred into individual graphite reduction reactors. The CO2 was reduced to graphite at 570 °C in the presence of an iron catalyst and a stoichiometric excess of hydrogen using procedures similar

1Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, Livermore, California 94551, USA. 2also at Department of Ocean Sciences, University of California-Santa Cruz, Santa Cruz, California 94596, USA. 3now at Department of Earth System Sciences, University of California-Irvine, Irvine, California 92967, USA.

75 76 T P Guilderson et al. to those described in Vogel et al. (1987). The graphite/iron samples produced were then pressed into aluminum target holders for subsequent AMS analysis. We have noticed an offset between acid-leached and non-acid-leached turbidite, whereby leached TIRI turbidite tends to be 100–200 yr older. This is undoubtedly due to the selective removal of finer carbonate particles (assumedly, coccolithophorids and planktonic foraminifera fragments). Similar results were observed as part of the TIRI intercomparison. The results presented within are from unleached TIRI turbidite. The prepared graphite targets were sputtered in a high-intensity cesium sputter source (Southon and Roberts 2000) with an equivalent 12C– current of 275–300 µA, which yields ~900 14C counts-second–1 on a modern carbon sample. After mass selection via the low-energy injector magnet, the negative ion beam (13C– or 14C– and molecular isobars) was injected into the accelerator (FN Tandem Van de Graff at CAMS), passed through a stripper foil, and on exiting the accelerator, was magnet- ically- and velocity-filtered, and subsequently measured in an off-axis faraday cup (13C4+) or analyzed in a gas ionization detector (14C4+) (Davis et al. 1990). The CAMS ion source sample wheel has slots for up to 64 targets and we normally load about 50 unknown samples in a routine sample wheel. Each wheel load is composed of a suite of about 10 aliquots of the primary (OX1) standard, plus secondary standards (OX2, ANU, TIRI B wood), the unknown samples, and blanks, and is bro- ken into several groups. In general, a group is composed of 10–14 unknowns with intervening and bracketing primary standards. Samples are analyzed in such a fashion that a single group is completely analyzed prior to proceeding onto the next. A group is analyzed repeatedly such that a suite of bracketing primary standards and the secondary standards are analyzed in conjunction with the unknown samples. For “high-precision” samples, a single group of unknowns is cycled through at least 5 times. During each cycle, an individual target is analyzed for either 30,000 14C events or 200 seconds, whichever comes first. The turbidite samples were interspersed throughout the sample wheel and were analyzed as unknown samples. Because of their low 14C concentration, the turbidite samples reached maximum time prior to obtaining 30,000 events. Cumulative 14C events for the TIRI turbidite results presented here are between 90,000 and 100,000 events. Raw data (14C/13C ratios) are normalized to the average of 6 bracketing aliquots of the primary standard for each pass through a sample group. Counting errors (primary standard and unknown) are propagated through the analysis and are assumed to be Gaussian (cf., Bevington and Robinson 1992). The average of the n-measurement cycles of each unknown is then determined and for the final error, the larger of the counting error or the external error of the n-cycles is chosen. CAMS 14C dates are based on 14C/13C atom ratios, not decay-counting, to obtain specific 14C activities. The algorithms used at CAMS (Southon, unpublished) are similar to those developed at Arizona (Donahue et al. 1990). Data are presented according to the conventions of Stuiver and Polach (1977). Calculations include a background subtraction and inclusion of background error based on 14C-free calcite determined on multiple aliquots of acid-leached calcite for each wheel of unknowns (cf., Brown and Southon 1997). The average background subtraction for the 8 runs presented here was FM = 0.0011 (n = 26). On any given wheel, the scatter of the backgrounds varied between 5– 35% and the appropriate scatter of the same blanks was used as the background uncertainty.

RESULTS The values obtained for the TIRI/FIRI turbidite samples are presented in Table 1, along with dates of AMS analysis and the preparation log-number (a unique log-number for samples prepared in our own graphite laboratory). These prep-log-numbers are reported because, historically, CAMS has not High-Precision Results on TIRI/FIRI Turbidite 77

Table 1 TIRI turbidite results from a high-precision 14C calibration project Date N-log # Fmoderna ±FM 14C age ±yr 18-Mar-01 N36555 0.10346 0.00041 18220 40 18-Mar-01 N36557 0.10406 0.00037 18180 30 18-Mar-01 N36691 0.10365 0.00041 18210 40 18-Mar-01 N36693 0.10380 0.00048 18200 40 2-May-01 N37419 0.10347 0.00049 18220 40 2-May-01 N37452 0.10452 0.00044 18140 40 5-Jun-01 N37431 0.10389 0.00042 18190 40 5-Jun-01 N37504 0.10364 0.00042 18210 40 5-Jun-01 N37443 0.10427 0.00042 18160 40 5-Jun-01 N37502 0.10430 0.00039 18160 30 5-Jul-01 N38795 0.10410 0.00047 18170 40 5-Jul-01 N38807 0.10399 0.00042 18180 40 17-Jul-01 N38814 0.10402 0.00045 18180 40 17-Jul-01 N38824 0.10446 0.00052 18150 50 17-Jul-01 N38831 0.10387 0.00045 18190 40 17-Jul-01 N38837 0.10382 0.00046 18200 40 6-Dec-01 N42165 0.10311 0.00033 18250 30 6-Dec-01 N42175 0.10355 0.00036 18215 30 6-Dec-01 N42177 0.10351 0.00041 18220 35 8-Jan-02 N42185 0.10330 0.00040 18235 35 8-Jan-02 N42193 0.10373 0.00041 18205 35 8-Jan-02 N42195 0.10391 0.00041 18190 35 12-Jun-02 N45680 0.10364 0.00061 18210 50 12-Jun-02 N45681 0.10358 0.00060 18215 50 12-Jun-02 N45682 0.10506 0.00059 18100 50 2-Sep-02 N47693 0.10297 0.00039 18260 35 2-Sep-02 N47705 0.10357 0.00046 18215 40 2-Sep-02 N47710 0.10380 0.00041 18195 35 aWe present the Fmodern results with 5 significant digits to allow readers to manipulate the data in a similar fashion as ourselves and to minimize propagation of round-off errors. assigned laboratory numbers to internal secondary or tertiary standards, backgrounds, test-samples, etc. As per Stuiver and Polach (1977), rather than using the reported ages in calculating a weighted mean, we report our results as the weighted mean of the measured Fraction Modern (FM) values of the TIRI turbidite samples. The weighted mean and weighted mean uncertainty (1σ) of the TIRI turbidite results are 0.10378 ± 0.00008 (n = 28), which equates to 18,199 ± 8 yr BP. Using the standard deviation of the TIRI/FIRI turbidite results, an external estimate of the uncertainty of the mean (SD/ √N) is 0.00008 (±8 yr). The distribution of the measurements and their quoted uncertainties yields a reduced chi-squared of 0.93 (Table 2).

DISCUSSION We find that the external fractional uncertainty in the 28 individual measurements is ~4‰ (1σ). This implies that our measurements on these 18,000-yr-old samples are reproducible at the ±35-yr level. Because of the vagaries of radioactive decay, this is a more appropriate characterization of the relative error than the common practice in the stable isotope community, which would take the standard 78 T P Guilderson et al.

Table 2 TIRI turbidite summary statistics (FM) Weighted mean 0.10378 Standard Deviation (1σ) 0.00008 Standard Dev in weighted mean (σ /√N) 0.00044 Error on wtd mean from individual measuring errors 0.00008 Reduced chi-squared 0.93 deviation of the reported Fmodern or “D14C” (in per mil) and call this the reproducibility of the samples. The reduced chi-squared value obtained for the results (χ2 = 0.93) shows that the scatter of the 28 measurements is consistent with the uncertainty estimates that have been derived from measurement error and background correction uncertainties. Taken with the above, this implies that our measurements are reproducible and precise within the quoted uncertainties. Graphical examination of the results (Figure 1) indicates that the distribution of the TIRI turbidites follow a Gaussian distribution. The scatter of these TIRI turbidite results is similar to recent results obtained on repeated analysis of a modern in-house coral standard (e.g., Guilderson et al. 2000). Our individual numbers basically agree with the TIRI and FIRI “consensus” values within the uncertainties of the individual measurements. However, our weighted-mean value of 18,199 ± 8 BP is inconsistent with (slightly older than) the consensus TIRI (18,155 ± 34 BP) and FIRI (18,173 ± 11 BP) results. Yet, it is worth noting that the difference between the age obtained in this study and the FIRI C “consensus” value is smaller than the differences between the FIRI turbidite values obtained when the contributing laboratories were grouped by technique (i.e., AMS-based results: 18,183 ± 13 BP, n = 34; GPC-based results: 18,229 ± 28 BP, n = 17; LSC-based results: 18,140 ± 25 BP, n = 34). Whether this is due to systematic but small offsets between turbidite aliquots measured by different laboratories, or to a biasing of the consensus results to younger ages by a subset of data incorporating over-optimistic background estimates, or to some other cause, is unclear. The increasing need for high-precision measurements in paleoclimate, archaeology, and modern- day 14C geophysical tracer studies places high demands on the capabilities of the 14C community to develop, validate, and maintain the ability to produce precise and accurate measurements that show appropriate levels of agreement. The utility of high-precision records produced at various 14C laboratories will depend significantly on the demonstrated ability of the laboratories to produce precise and accurate measurements that show appropriate levels of agreement between the high-precision laboratories. The TIRI and FIRI sample suite provide a basis for such laboratory intercomparisons but it is apparent that accurate and precise ages of these materials may yet need to be determined. Further high-precision measurements of consistency samples, such at those provided by FIRI and TIRI, need to be made in order to ascertain which laboratories are capable of the reproducibility necessary for calibration quality measurements and to determine systematic laboratory offsets that may need to be accounted for when compiling results from different laboratories.

SUMMARY We have made reproducible measurements of an 18,000-BP sample over an 18-month period that scatter with a standard deviation equivalent to a ~4‰ fractional error. The scatter of these measurements is consistent with the counting-statistics-based estimates of the uncertainties in our measurements (i.e., we can do 4‰ on 18ka BP samples and see essentially nothing but counting statistics). Our measurements are accurate and consistent with the current knowledge of the 14C content of the sample. Other laboratories should endeavor to perform reasonably high-precision analysis of the High-Precision Results on TIRI/FIRI Turbidite 79

10 A

Observations 2

0 .1015 .1025 .1035 .1045 .1055 .1065 Fraction Modern 12

10 B

4 Observations 2

0 17950 18050 18150 18250 18350 14C yr BP Figure 1 Distribution of TIRI turbidite results as a function of fraction modern (A) and conventional 14C age. Respective curves are predicted Gaussian distributions determined from the mean and standard deviation of the turbidite results.

TIRI/FIRI sample suite so that a more accurate age determination of these materials is provided to the community. Additional sample materials, such as those provided by the TIRI/FIRI intercomparison projects, will be invaluable to constrain the accuracy and reproducibility of any individual suite of high-precision data. 80 T P Guilderson et al.

ACKNOWLEDGEMENTS Special thanks to J Westbrook and P Zermeno for pressing the graphite targets made by TG. Discus- sions with Paula Reimer, Gerry McCormac, and Marian Scott are gratefully acknowledged. We thank the Radiocarbon editorial staff and George Burr for reviewing this manuscript. 14C analyses were performed under the auspices of the US Department of Energy by the University of California Lawrence Livermore National Laboratory (contract W-7405-Eng-48).

REFERENCES Bevington PR, Robinson DK. 1992. 2nd edition. Data Hughen KA, Overpeck JT, Lehman SJ, Kashgarian M, Reduction and Error Analysis for the Physical Sci- Southon JR. 1998. A new C-14 calibration data set for ences. Boston: McGraw-Hill. 328 p. the last deglaciation based on marine varves. Radio- Brown TA, Southon JR, Roberts MR. 2000. Ion-source carbon 40:483–94. modeling and improved performance of the CAMS Hughen KA, Southon JR, Lehman SJ, Overpeck JT. high-intensity Cs-sputter ion source. Nuclear Instru- 2000. Synchronous radiocarbon and climate shifts ments and Methods in Physics Research, Section B, during the last deglaciation. Science 290:1951–4. 172:344–9. McCormac FG, Hogg AG, Higham TFG, Lynch-Stieglitz Brown TA, Southon JR. 1997. Corrections for contami- J, Broecker WS, Baillie MGL, Palmer J, Xiong L, nation background in AMS 14C measurements. Nu- Pilcher JR, Brown D, Hoper S. 1998. Temporal varia- clear Instruments and Methods in Physics Research, tion in the interhemispheric C-14 offset. Geophysical Section B, 123:208–13. Research Letters 25:1321–4. Davis JC, Proctor ID, Southon JR, Caffee MW, Heikki- Southon JR. The calculation of 14C ages from AMS 14C/ nen DW, Roberts ML, Moore TL, Turteltaub KW, 13C ratio measurements. Unpublished document. Nelson DE, Loyd DH, Vogel JS. 1990. LLNL/UC Southon JR, MR Roberts. Ten years of sourcery at cams/ AMS facility and research program. Nuclear Instru- LLNL-evolution of a Cs ion source. Nuclear Instru- ments and Methods in Physics Research, Section B, ments and Methods in Physics Research, Section B, 52: 269–72. 172:257–61. Donahue DJ, TW Linick, AJT Jull. 1990. Isotope-ratio Stuiver M, Polach H. 1977. Discussion of reporting of and background corrections for accelerator mass spec- 14C data. Radiocarbon 19:355–63. trometry radiocarbon measurements. Radiocarbon Vogel JS, Southon JR, Nelson DE. 1987. Catalyst and 32:135–42. binder effects in the use of filamentous graphite for Fairbanks RG, Mortlock RM, Guilderson T, Rubenstone AMS. Nuclear Instruments and Methods in Physics JL, Chiu T-C, Hubbard DK. 2001. Radiocarbon cali- Research, Section B, 29:50–6. bration via U/Th/Pa on pristine corals. Geological So- Weiss H, deLillis F, de Moulins D, Eidem J, Guilderson ciety of America 33(6):22. T, Kasten U, Larsen T, Mori L, Ristvet L, Rova E, Guilderson TP, Schrag DP, Goddard E, Kashgarian M, Wetterstrom W. 2002. Revising the contours of history Wellington GM, Linsley BK. 2000. Southwest sub- at Tell Leilan. Annales Archeologiques Arabes Syri- tropical Pacific surface water radiocarbon in a high- ennes, Cinquantenaire. resolution coral record. Radiocarbon 42:249–56. RADIOCARBON, Vol 45, Nr 1, 2003, p 81–89 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

RADIOCARBON CALIBRATION FOR JAPANESE WOOD SAMPLES

Minoru Sakamoto1 • Mineo Imamura1,2 • Johannes van der Plicht3 • Takumi Mitsutani4 • Makoto Sahara1,5

ABSTRACT. The radiocarbon content of Japanese cedars was measured by accelerator mass spectrometry for decadal tree- ring samples from the period of 240 BC to AD 900. Conventional gas counting was also used for part of the samples. The data were compared with the INTCAL98 calibration curve (Stuiver et al. 1998). The results indicate that the difference in atmospheric 14C between Japan and North America or Europe is negligible at this period, less than 18 14C yr using an average of 50 yr. However, in the period of about AD 100 to about AD 200, we cannot exclude the possibility of a deviation of the order of 30 to 40 14C yr to the older ages.

INTRODUCTION Precise calibration data are available for the atmospheric 14C record of the past, enabling us to estimate true ages with much more confidence than before. The combination of 14C dates and dendrochronology can lead to an exact age determination. Moreover, accurate dates—up to about 10–20 calibrated yr—can be obtained by means of the so-called wiggle-matching technique. Therefore, precise dating by 14C is of increasing interest for archaeologists. Precise knowledge on the local effect of the atmospheric 14C content is becoming more important for future perspectives of precise 14C dating. Thus far, conversion of 14C dates to calendar yr has been based on 14C calibration curves obtained from North American and European trees for Japanese samples. No calibration curve for Asian trees is available. It is important to examine possible regional deviation from INTCAL98 and investigate the extent of the regional effect in 14C calibration. By doing this, we can safely apply the INTCAL98 calibration curve to the Japanese samples. The aim of this research is to obtain precise 14C dates for dendrochronologically-dated woods, particularly for the Yayoi and Kofun periods, which are described below. For archaeological research in Japan, the period of several centuries around AD 1 has been of great interest because in this period rapid changes took place in many cultural and social aspects. The first half of the period corresponds to the prehistoric era (Yayoi period, 4th century BC to 3rd century AD) during which paddy rice farming has spread over the islands and the use of bronze and iron started. The latter half corresponds to the protohistoric era (Kofun period, 3rd century AD to 6th century AD), which is characterized by huge burial mounds and the establishment of the first state in Japan. It is also conceived that cultural exchange was quite active between the East Asian continent and the Japanese archipelago. The absolute chronology of these periods is controversial, since recent dendrochronological dates obtained for several Yayoi archaeological sites show a significant discrepancy from the archaeological chronology that had been widely accepted (Mitsutani 2002). Therefore, precise dating has been of great concern for many archaeologists in Japan as well. Recent studies show that 14C dating within a few tens of yr is possible using the wiggle-matching technique, as mentioned above. The use of the universal calibration curves for Japanese wood samples should be suitable for practical use, however, it is necessary to prove that the curves are correct.

1National Museum of Japanese History, 117 Jonai-cho, Sakura-shi, Chiba 285-8502, Japan. Email: [email protected]. 2Department of Japanese History, Graduate University for Advanced Studies, 117 Jonai-cho, Sakura-shi, Chiba 285-8502, Japan. 3Center for Isotope Research, University of Groningen, NL-9747 AG Groningen, the Netherlands. 4Nara National Cultural Properties Research Institute, 2-9-1 Nijo-cho, Nara-shi, Nara 630-8002, Japan. 5Deceased July 10, 2002.

81 82 M Sakamoto et al.

We have measured the 14C dates of decadal tree rings using 4 Japanese wood specimens that have been dendro-dated. By comparing 14C dates with the INTCAL98 calibration curve, we discuss in detail the applicability of INTCAL98 to the precise 14C dating of the pre- and proto-historic periods in Japan. This produces the first contribution to a calibration curve for Japan. Because of the limited amount of wood available, most of the measurements were done by AMS. Eleven measurements were done by the conventional method.

DESCRIPTIVE BACKGROUND

Sample source and tree-ring dating Four Japanese cedar (Cyptomeria japonica) specimens—HK, MT, AH1, and AH3—were selected for this study. HK was a bogwood unearthed at Hakone, Kanagawa Prefecture, central Honshu Island located about 100 km west of Tokyo. The ring patterns of HK spanned 453 yr. MT was also a bogwood unearthed at Miyata-mura village, Nagano Prefecture, central Honshu Island, with total ring patterns of 372 yr. AH1 and AH3 were individual piles excavated at the Hotta-no-saku archaeological site in Akita Prefecture, northern Honshu Island. The locations of these samples are shown in Figure 1.

130°E 140°E

40°N 40°N AH1, AH3

MT Honshu island JAPAN

30°N 30°N

130°E 140°E

Figure 1 Map of sample locations

Dendrochronological determination of the absolute age for these specimens was performed by comparing the standardized ring patterns with the master chronology for Japanese cedar developed by the Nara National Culture Properties Research Institute (1990). A master standard chronology for a Japanese cedar has been established for tree-ring ages back to 1313 BC. Each specimen was cut into decadal pieces according to its dendro-date. Although the ring width varied, we assumed that each decadal piece represents the mean 14C concentration in that time Calibration for Japanese Wood Samples 83 period. Thirty-two samples were obtained from HK corresponding to the age of 240 BC to AD 200, 30 samples from MT (AD 331 to AD 630), 11 samples from AH1 (AD 692 to AD 801), and 24 samples from AH3 (AD 661 to AD 900), respectively. Preliminary measurement using the HK specimen for 201 BC to AD 60 was reported by Imamura et al. (1998). The surface portion of the sample was removed to prevent possible contamination with modern carbon, and each decadal sample piece was pulverized into fine fiber tips of about 0.1 × 1 mm size using a mill (an electric blender).

14C measurements

Accelerator Mass Spectrometry (AMS) (lab. code GrA) The HK, MT, and AH3 samples were measured for 14C using AMS. About 100 mg of the sample fiber was treated by acid-alkali-acid (in each step, the sample was reacted 5 times with 1.2N HCl or 1.2N NaOH for 1 hr), followed by removal of lignin with Cl2 (using NaClO2 and HCl) and most β- and γ- cellulose with a 17.5% NaOH solution. The purified sample, mainly consisting of α-cellulose, was then neutralized, washed with pure H2O, filtrated, and dried. Several mg of cellulose were taken in a Vycor-glass tube together with few hundred mg of CuO (organic carbon analysis grade: Wako Chemical Co.) and sealed off from the vacuum system. The sample tube was then heated at 850 ºC for 2 hr to completely oxidize the cellulose. The obtained CO2 was transferred to the high-vacuum CO2 purification systems (at the National Museum of Jap- anese History for HK and AH3, and at the University of Groningen for MT) and purified cryogenically. The pressure of the purified CO2 was measured by a gauge and divided into 2 breakseals. Graphite targets were prepared in Groningen and the 14C/12C and 13C/12C ratios were measured with the Groningen HVEE AMS system (van der Plicht et al. 2000). 14C dates were determined from the 14C/12C ratios after normalizing the isotopic ratio of 13C/12C to δ13C = –25.0 per mil.

Conventional 14C measurements (lab. code GrN) The AH1 samples were measured for 14C in the Groningen conventional laboratory. Several g of the sample fiber were treated by the acid-alkali-acid method. Treated fiber was combusted into CO2 and purified. Obtained CO2 was collected in a metal cylinder, introduced into a proportional counter, and measured for 14C β-rays radiation (Mook and Streurman 1983; van der Plicht et al. 1992). Also, here the 14C dates are reported including a correction for fractionation to δ13C = –25.0 per mil.

DISCUSSION The results of the 14C measurements for each sample specimen are shown in Tables 2–5. The data include our preliminary report on the 200–60 BC samples for HK (Imamura et al. 1998). The errors include the statistics of the 14C counts and uncertainties in the 13C/12C and 14C/12C ratios of the standard and blank targets. The extent of modern carbon contamination was of the order of 0.1% or less, which is negligible. Uncertainties of original 14C standards are not included. In Figure 2, the results are compared with the calibration curve of INTCAL98 (Stuiver et al. 1998). Error bars represent a 68% (1σ) uncertainty. From the data plotted in Figure 2, it is shown that almost all the data are in good agreement (within 2σ) with the INTCAL98 calibration curve. There may be an exception in the period of around AD 100–200, which may show a significant deviation toward the older ages and will be discussed later in more detail. 84 M Sakamoto et al.

2300 INTCAL98 HK (240 BC - AD 200) MT (AD 331 - AD 630) 2100 AH1 (AD 692 - AD 801) AH3 (AD 661 - AD 900)

1900

1700 C age (BP) 14 1500

1300

1100 200 BC/AD 200 400 600 800 Calendar Year Figure 2 14C data for Japanese cedars compared with INTCAL98

Table 1 Source of the tree-ring sample specimens Sample code Locality Latitude Longitude Tree-ring age HK Hakone-machi Town, central 35°10′N 139°05′E 240 BC–AD 200 Honshu Island, Japan MT Miyata-mura Village, central 35°45′N 137°55′E AD 331–AD 630 Honshu Island, Japan AH1 Hotta-no-saku archaeological site, 39°30′N 140°35′E AD 692–AD 801 northern Honshu Island, Japan AH3 Hotta-no-saku archaeological site, 39°30′N 140°35′E AD 661–AD 900 northern Honshu Island, Japan

The significance of deviation for each datum from the INTCAL98 can be given by the formula:

Tt( ) − Tt( ) ∆=Z sINTCAL98 22 σσsINTCAL98+

14 where ∆Z is the normalized difference, T(t)S and T(t)INTCAL98 are the C ages of the decadal tree-ring sample and the corresponding INTCAL98 curve for a calibrated age, t, and σs, σINTCAL98 are their respective uncertainties. The ∆Z values are calculated for the individual dataset of each specimen. The distribution of cases is plotted in Figure 3 as a function of ∆Z. The total number of the case is normalized to 1. Figure 3 shows the distribution for each sample specimen is nearly a typical normal distribution and the average of the distribution is very close to zero. This is what is expected when there are no biases in the measurements and no significant regional effect for the whole period investigated. By calculating a best fit to a normal distribution, the shifts are calculated to be 0.42, –0.23, Calibration for Japanese Wood Samples 85

Table 2 Results of 14C measurements (AMS) for the HK series Lab. Code Sample name Tree-ring age 14C age (BP) GrA-14700 HK-235B 240 BC – 231 BC 2225 ± 40 GrA-14702 HK-225B 230 BC – 221 BC 2255 ± 40 GrA-14703 HK-215B 220 BC – 211 BC 2240 ± 40 GrA-14704 HK-205B 210 BC – 201 BC 2235 ± 40 GrA-9332 HK-195B 200 BC – 191 BC 2125 ± 45 GrA-9326 HK-185B 190 BC – 181 BC 2160 ± 45 GrA-8085 HK-175B 180 BC – 171 BC 2150 ± 75 GrA-9323 HK-165B 170 BC – 161 BC 2090 ± 45 GrA-9330 HK-165B 170 BC – 161 BC 2180 ± 45 GrA-9328 HK-155B 160 BC – 151 BC 2130 ± 45 GrA-8082 HK-145B 150 BC – 141 BC 2130 ± 65 GrA-9419 HK-145B 150 BC – 141 BC 2080 ± 40 GrA-8081 HK-135B 140 BC – 131 BC 2150 ± 65 GrA-9322 HK-125B 130 BC – 121 BC 2085 ± 45 GrA-9327 HK-115B 120 BC – 111 BC 2035 ± 45 GrA-9316 HK-105B 110 BC – 101 BC 2080 ± 45 GrA-9317 HK-105B 110 BC – 101 BC 2120 ± 45 GrA-9318 HK-095B 100 BC – 91 BC 2085 ± 45 GrA-9321 HK-095B 100 BC – 91 BC 2050 ± 45 GrA-8084 HK-085B 90 BC – 81 BC 2055 ± 65 GrA-14705 HK-075B 80 BC – 71 BC 2095 ± 40 GrA-14707 HK-065B 70 BC – 61 BC 2155 ± 40 GrA-14709 HK-055B 60 BC – 51 BC 2095 ± 40 GrA-14689 HK-045B 50 BC – 41 BC 2060 ± 40 GrA-14688 HK-035B 40 BC – 31 BC 2010 ± 40 GrA-14627 HK-025B 30 BC – 21 BC 2005 ± 40 GrA-14628 HK-015B 20 BC – 11 BC 2030 ± 40 GrA-14629 HK-005B 10 BC – 1 BC 2065 ± 40 GrA-14630 HK-005A AD 1 – AD 10 1965 ± 40 GrA-14632 HK-015A AD 11 – AD 20 2000 ± 40 GrA-14684 HK-025A AD 21 – AD 30 2030 ± 40 GrA-14685 HK-035A AD 31 – AD 40 2010 ± 40 GrA-14690 HK-045A AD 41 – AD 50 1975 ± 40 GrA-14691 HK-055A AD 51 – AD 60 1985 ± 40 GrA-14693 HK-065A AD 61 – AD 70 1925 ± 40 GrA-14694 HK-075A AD 71 – AD 80 1895 ± 40 GrA-14695 HK-085A AD 81 – AD 90 1935 ± 40 GrA-14698 HK-095A AD 91 – AD 100 1950 ± 40 GrA-14699 HK-105A AD 101 – AD 110 1950 ± 40 GrA-14710 HK-115A AD 111 – AD 120 1985 ± 40 GrA-14712 HK-125A AD 121 – AD 130 1910 ± 40 GrA-14713 HK-135A AD 131 – AD 140 1900 ± 40 GrA-14714 HK-145A AD 141 – AD 150 1885 ± 40 GrA-14715 HK-155A AD 151 – AD 160 1885 ± 40 GrA-14717 HK-165A AD 161 – AD 170 1810 ± 40 GrA-14718 HK-175A AD 171 – AD 180 1870 ± 40 GrA-14720 HK-185A AD 181 – AD 190 1915 ± 40 GrA-14722 HK-195A AD 191 – AD 200 1920 ± 40 86 M Sakamoto et al.

Table 3 Results of 14C measurements (AMS) for the MT series Lab. code Sample name Tree-ring age 14C age (BP) GrA-15263 MT-335 AD 331 – AD 340 1755 ± 50 GrA-15264 MT-345 AD 341 – AD 350 1620 ± 50 GrA-15265 MT-355 AD 351 – AD 360 1730 ± 50 GrA-15266 MT-365 AD 361 – AD 370 1640 ± 50 GrA-15268 MT-375 AD 371 – AD 380 1725 ± 50 GrA-15269 MT-385 AD 381 – AD 390 1675 ± 50 GrA-15270 MT-395 AD 391 – AD 400 1640 ± 50 GrA-15273 MT-405 AD 401 – AD 410 1580 ± 50 GrA-15320 MT-415 AD 411 – AD 420 1660 ± 50 GrA-15275 MT-425 AD 421 – AD 430 1580 ± 50 GrA-15276 MT-435 AD 431 – AD 440 1655 ± 50 GrA-15278 MT-445 AD 441 – AD 450 1595 ± 50 GrA-15279 MT-455 AD 451 – AD 460 1640 ± 50 GrA-15280 MT-465 AD 461 – AD 470 1590 ± 50 GrA-15283 MT-475 AD 471 – AD 480 1645 ± 50 GrA-15287 MT-485 AD 481 – AD 490 1520 ± 50 GrA-15289 MT-495 AD 491 – AD 500 1570 ± 50 GrA-15291 MT-505 AD 501 – AD 510 1555 ± 50 GrA-24950 MT-515 AD 511 – AD 520 1480 ± 50 GrA-15293 MT-525 AD 521 – AD 530 1535 ± 50 GrA-15294 MT-535 AD 531 – AD 540 1520 ± 50 GrA-15296 MT-545 AD 541 – AD 550 1445 ± 50 GrA-15298 MT-555 AD 551 – AD 560 1480 ± 50 GrA-15299 MT-565 AD 561 – AD 570 1480 ± 50 GrA-15301 MT-575 AD 571 – AD 580 1385 ± 50 GrA-15302 MT-585 AD 581 – AD 590 1540 ± 50 GrA-15303 MT-595 AD 591 – AD 600 1485 ± 50 GrA-15304 MT-605 AD 601 – AD 610 1495 ± 50 GrA-15306 MT-615 AD 611 – AD 620 1490 ± 50 GrA-15307 MT-625 AD 621 – AD 630 1350 ± 50

Table 4 Results of 14C measurements (conventional) for the AH1 series Lab. code Sample name Tree-ring age δ13C (‰) 14C age (BP) GrN-25080 AH1-696 AD 692 – AD 701 –26.35 1287 ± 25 GrN-25081 AH1-706 AD 702 – AD 711 –26.18 1284 ± 25 GrN-25082 AH1-716 AD 712 – AD 721 –26.42 1255 ± 25 GrN-25083 AH1-726 AD 722 – AD 731 –26.85 1287 ± 26 GrN-25084 AH1-736 AD 732 – AD 741 –26.57 1286 ± 23 GrN-25085 AH1-746 AD 742 – AD 751 –26.37 1313 ± 26 GrN-25086 AH1-756 AD 752 – AD 761 –26.56 1317 ± 26 GrN-25087 AH1-766 AD 762 – AD 771 –26.34 1301 ± 26 GrN-25088 AH1-776 AD 772 – AD 781 –25.58 1255 ± 43 GrN-25089 AH1-786 AD 782 – AD 791 –25.70 1228 ± 31 GrN-25090 AH1-796 AD 792 – AD 801 –25.37 1227 ± 28 Calibration for Japanese Wood Samples 87

Table 5 Results of 14C measurements (AMS) for the AH3 series Lab. Code Sample name Tree-ring age 14C age (BP) GrA-14997 AH3-665 AD 661 – AD 670 1340 ± 60 GrA-14998 AH3-675 AD 671 – AD 680 1320 ± 60 GrA-14999 AH3-685 AD 681 – AD 690 1250 ± 60 GrA-15000 AH3-695 AD 691 – AD 700 1240 ± 60 GrA-15002 AH3-705 AD 701 – AD 710 1230 ± 60 GrA-15003 AH3-715 AD 711 – AD 720 1250 ± 60 GrA-15004 AH3-725 AD 721 – AD 730 1270 ± 60 GrA-15007 AH3-735 AD 731 – AD 740 1240 ± 60 GrA-15008 AH3-745 AD 741 – AD 750 1280 ± 60 GrA-15009 AH3-755 AD 751 – AD 760 1320 ± 60 GrA-15082 AH3-765 AD 761 – AD 770 1310 ± 45 GrA-15084 AH3-775 AD 771 – AD 780 1225 ± 45 GrA-15086 AH3-785 AD 781 – AD 790 1230 ± 45 GrA-15095 AH3-795 AD 791 – AD 800 1245 ± 45 GrA-15139 AH3-805 AD 801 – AD 810 1215 ± 40 GrA-15140 AH3-815 AD 811 – AD 820 1115 ± 40 GrA-15143 AH3-825 AD 821 – AD 830 1200 ± 40 GrA-15144 AH3-835 AD 831 – AD 840 1210 ± 40 GrA-15146 AH3-845 AD 841 – AD 850 1190 ± 40 GrA-15153 AH3-855 AD 851 – AD 860 1235 ± 40 GrA-15150 AH3-865 AD 861 – AD 870 1225 ± 40 GrA-15151 AH3-875 AD 871 – AD 880 1235 ± 40 GrA-15152 AH3-885 AD 881 – AD 890 1165 ± 40 GrA-15155 AH3-895 AD 891 – AD 900 1185 ± 40

HK (240 BC - AD 200) MT (AD 330 - AD 630) 30 AH3 (AD 660 - AD 900) Average

Distribution (%/0.5unit) 10

0 -3 -2 -1 0 1 2 3 ∆Z Figure 3 Difference between 14C data for Japanese cedars and INTCAL98 88 M Sakamoto et al. and 0.16 for HK, MT and AH3, respectively. The data of AH1 are too small to show a meaningful distribution. If we take a typical error of each determination, the shifts are 18, –10, and 6 14C yr, respectively. The wiggle-matched date of each specimen is calculated using all the data by comparing with the INTCAL98 calibration curve. The calculated dates are in very good agreement with the dendrochronological dates determined for each specimen. Although it is known that the calculated age should be less influenced by the regional effect (Bronk Ramsey et al. 2001), the observations above also enforce the correctness of using INTCAL98 calibration curve for Japanese wood samples. The above conclusion is based on the data set of each sample specimen of a rather long period. To search the detailed structure of the regional effect, 50 yr of averages of the differences between the data and INTCAL98 are calculated and plotted in Figure 4. The figure shows the difference is within the 2σ uncertainties except for the values around AD 100 to AD 200, indicating absence of regional effect that exceeds 20 14C yr in the time scale of 50 calendar yr.

200

100 C years) 14

( 0 INTCAL98 -T s

T -100 50 years moving average ±1s HK (240 BC - AD 200) -200 200 100 BC/AD 100 200 Calendar Year Figure 4 Difference of 14C ages between Japanese cedar and INTCAL98

For the data of AD 100 to AD 200, we possibly observe a regional or local effect. The 50-yr averages for these periods certainly deviate more than the 2σ uncertainties from those of the INTCAL98. At present, we consider 2 possibilities: that the regional effect exists in this period for Japanese islands, or that the effect is only local due to the volcanic activities from a nearby volcano (Mt. Fuji for example) and restricted to the area around the Fuji-Hakone National Park in Japan where the HK sample was taken. We plan to measure another tree-ring sample of the same period but taken from a different area of Japan.

CONCLUSIONS From the present study, it is clear that the differences in atmospheric 14C concentrations between Japan and North America or Europe are negligible or small during the period of 270 BC to AD 900, suggesting that there is little disturbance of 14C in the atmosphere caused by the ocean near the Jap- anese archipelago. This is likely because the air mass moves eastwards at the mid-latitude and Calibration for Japanese Wood Samples 89

reaches Japan with little influence of the ocean. However, we found a possible deviation in the period of about AD 100 to AD 200, which requires future investigation.

ACKNOWLEDGMENTS We are indebted to Professors Hiroyuki Kitagawa and Toshio Nakamura for useful discussions and comments. J van der Plicht is indebted to the Japan Society of Promotion of Science (S02027) for their support and fellowship. The work is supported by a Grant-in-Aid (No. 09301017) of the Min- istry of Education, Science, Culture, and Sports in Japan.

REFERENCES Imamura M, Sakamoto M, Shiraishi T, Sahara M, Naka- the Nara National Cultural Research Institute 48. In mura T, Mitsutani T, van der Plicht J. 1998. Radiocar- Japanese with summary in English. bon age calibration for Japanese wood samples: wig- Ramsey B, van der Plicht J, Weninger B. 2001. “Wiggle gle-matching analysis for a test specimen. In: Evin J, matching” radiocarbon dates. Radiocarbon 43(2A): Oberlin C, Daugas JF, Salles JF, editors. Proceedings 381–9. of the 3rd International Symposium on 14C and Ar- Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, chaeology, Lyon. Revue d’Archeometrie 23:79–82. Hughen KA, Kromer B, McCormac G, van der Plicht Mitsutani T. 2002. Dendrochronology in Japan and its J, Spurk M. 1998. INTCAL98 Radiocarbon Age Cal- applications to archaeology. Paper presented at the 6th ibration, 24,000–0 cal BP. Radiocarbon 40(3):1041– International Conference on Dendrochronology. Que- 83. bec, Canada, 22–27 August. van der Plicht J, Streurman HJ, Schreuder GR. 1992. A Mook WG, Streurman HJ. 1983. Physical and chemical new data acquisition system for the Groningen aspects of radiocarbon dating. In: Mook WG, Water- counters. Radiocarbon 34(3):500–5. bolk HT, editors. Proceedings of the 2nd International van der Plicht J, Wijma S, Aerts AT, Pertuisot MH, Symposium on 14C and Archaeology. Groningen: Meijer HA.J. 2000. Status report: The Groningen PACT Publishers 8:31–55. AMS facility. Nuclear Instruments and Methods, Sec- Nara National Cultural Properties Research Institute. tion B 172(1–4):58–65. 1990. Dendrochronology in Japan. Research Report of RADIOCARBON, Vol 45, Nr 1, 2003, p 91–99 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

DECADAL TIMESCALE SHIFT IN THE 14C RECORD OF A CENTRAL EQUATORIAL PACIFIC CORAL

A G Grottoli1,2 • S T Gille3 • E R M Druffel4 • R B Dunbar5

ABSTRACT. Coral skeletal radiocarbon records reflect seawater ∆14C and are useful for reconstructing the history of water mass movement and ventilation in the tropical oceans. Here, we reconstructed the inter-annual variability in central equatorial Pacific surface water ∆14C from 1922–1956 using near-monthly 14C measurements in a Porites sp. coral skeleton (FI5A) from the windward side of Fanning Island (3°54′32″N, 159°18′88″W). The most pronounced feature in this record is a large, positive shift in the ∆14C between 1947 and 1956 that coincides with the switch of the Pacific Decadal Oscillation (PDO) from a positive to a negative phase in the mid-1940s. Although the absolute ∆14C values from 1950–1955 in FI5A differ from the ∆14C values of another coral core collected from the opposite side of the island, both records show a large, positive shift in their ∆14C records at that time. The relative increase in the ∆14C of each record is consistent with the premise that a common mechanism is controlling the ∆14C records within each coral record. Overall, the Fanning ∆14C data support the notion that a significant amount of subtropical seawater is arriving at the Equator, but does not allow us to determine the mechanism for its transport.

INTRODUCTION Coral proxy records offer a viable means of recovering pre-instrumental climate and oceanographic information. Scleractinian corals deposit a calcium carbonate skeleton (aragonite) in distinct, annual bands and can grow for several hundred yr. X-radiographic analysis of a thin slab of coral skeleton typically reveals alternating light and dark bands, each pair of which represents 1 yr of coral growth (e.g., Weber et al. 1975; Hudson et al. 1976; Barnes and Lough 1993). Corals incorporate the dissolved inorganic carbon (DIC) of the surrounding seawater as aragonite, a crystalline form of calcium carbonate, into their exoskeleton. Coral radiocarbon (∆14C) values have been shown to reflect the ∆14C of seawater DIC (Druffel 1997) and are useful for determining the history of water mass movement and ventilation in the subtropical and tropical oceans (e.g., Nozaki et al. 1978; Druffel 1987; Guilderson and Schrag 1998a; Druffel et al. 2001). 14C is produced naturally in the stratosphere and was also created anthropogen- ically as a result of thermonuclear weapons explosions in the stratosphere in the late 1950s and early 1960s. ∆14C in seawater DIC is a natural tracer of upper-ocean circulation. Ocean waters from below the thermocline have lower ∆14C values than surface waters because they are isolated from the atmosphere, the source of 14C. Therefore, surface concentrations of seawater ∆14C are sensitive to upwelling and vertical mixing. Prior to 1957, non-polar surface water 14C values ranged from –38‰ in mid-gyre regions to –72‰ in the eastern equatorial Pacific, where upwelling is very strong (Table 1). Here, we reconstructed the pre-bomb record of ∆14C in surface waters of the central equatorial Pacific from 1922–1956 from Fanning Island and found a profound change in 14C that correlated with a shift in the climate of the North Pacific Ocean.

1Department of Earth and Environmental Science, University of Pennsylvania, 240 South 33rd Street, Philadelphia, Pennsylvania 19104-6316, USA. 2Corresponding author. Email: [email protected]. 3Scripps Institution of Oceanography and Department of Mechanical and Aerospace Engineering, University of California- San Diego, La Jolla, California 92093-0230, USA. Email: [email protected]. 4Department of Earth System Science, University of California-Irvine, Irvine, California 92697-3100, USA. Email: [email protected]. 5Department of Geological and Environmental Science, Stanford University, Stanford, California 94305-2115, USA. Email: [email protected].

91 92 A G Grottoli et al.

METHODS FI5A is a longitudinal slab cut from a 75-yr-long coral core taken from a live head spanning 1922– 1997. The coral core was collected at 11 m depth in October 1997 outside the lagoon on the eastern (windward) side of Fanning Island (3°54′32″N, 159°18′88″W). The core length is 126.0 cm with an average yearly growth of 1.7 cm. This study focuses on the natural variation in coral ∆14C in the pre- bomb (pre-1955) period. We chose to focus our study on the pre-bomb period, when ambient atmospheric ∆14C levels are believed to be relatively constant. Samples were collected for 14C analysis every millimeter by milling a 14-mm-wide by 4-mm-deep trough along the major axis of growth using a Dremel tool with a diamond-tipped drill bit. Care was taken to follow the curvature of the horizontal growth lines within each sample in order to minimize chronological smearing. For 1922–1956, the 14C value was measured on every other sample, yielding an average of 8 sub-annual samples per yr. Approximately 50 mg of coral was acidified over- night at room temperature in a small glass 50 ml volume acidification vessel under vacuum and a 1.6 ml subsample of the resultant CO2 gas was reduced with hydrogen gas on an iron metal catalyst to 13 produce a 0.8 mg graphite target (Vogel et al.). δ C was measured on a second subsample of CO2 gas using a Finnigan MAT 251 at NOSAMS (δ13C = the ‰ deviation of 13C:12C relative to the v- Peedee Belemnitel Limestone Standard). The 14C content of the graphite was measured using accelerator mass spectrometry (AMS) techniques and reported as ∆14C (the per mil deviation of 14C/12C of the sample relative to that of the 95% Oxalic Acid-1 standard) (Stuiver and Polach 1977). All ∆14C values were corrected for fractionation to a δ13C of –25‰. One-third (104 samples) of the AMS measurements were performed at NOSAMS (National Ocean Sciences AMS Facility at the Woods Hole Oceanographic Institution) with an uncertainty of ±5.0–7.0‰. Two-thirds (209 samples) of the AMS measurements were made at CAMS (Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratories) with an uncertainty of ±3.0‰ or less. Numerous analyses of an internal laboratory reference coral standard averaged –59.0 ± 5.4‰ (1 standard deviation, 14 samples) at NOSAMS and averaged –60.3 ± 3.6‰ (1 standard deviation, 20 samples) at CAMS. The average ∆14C values of the internal laboratory reference standard did not significantly differ between laboratories (t-test: t = 0.86, degrees of freedom = 32, p = 0.4), though the precision of measurements at CAMS is higher than that at NOSAMS.

RESULTS Figure 1 shows ∆14C values measured at roughly 6-week intervals for the 35-yr pre-bomb period of the record from 1922–1956. Overall ∆14C values range from –71‰ in 1944 to –17‰ in 1951. Pre- 1947 values range from –71‰ to –39‰ and post-1947 values range from –52‰ to –17‰. Seasonal- to-interannual variations in the ∆14C record are not statistically detectable by spectral analysis. The most pronounced feature in this record is the noticeable positive shift in the ∆14C beginning in 1947 that lasts until 1956. The overall yearly average ∆14C increased by 17‰ from –55 ± 5‰ (1 standard deviation, sample size = 217) to –38 ± 8‰ (1 standard deviation, sample size = 78) after 1947. Dif- ferences in the ∆14C measurement precision between NOSAMS and CAMS do not interfere with our ability to interpret our results because the 17‰ shift in Fanning coral ∆14C is much greater than the reported measurement errors. An element of complexity enters our interpretation of the ∆14C when we compare the results for FI5A with those from another coral from the leeward side of Fanning Island (CTFN). An annual resolution ∆14C time series is available from the CTFN coral, located 11 km to the west of the FI5A site, on the opposite side (leeward side) of the island at ~10 m depth (3°55′N; 159°24′W) (Druffel 1987). Both FI5A and CTFN ∆14C records show an increase in the ∆14C in the 1940s to 1950s, but Shift in 14C record of an equatorial Pacific coral 93

Figure 1 Near-monthly skeletal ∆14C (‰) of Fanning Island Porites coral (FI5A) from 1922–1956. ∆14C averaged –55‰ from 1922–1947and –38‰ from 1947–1956, as indicated by dashed lines. Individual ∆14C values measured at CAMS (•) and NOSAMS (•) are plotted and connected with a solid black line. Two bulk annual ∆14C measurements for 1956 and 1959 (*) are plotted. Areas of overlap between 2 adjacent drilling paths (light gray lines) closely match and are barely visible. differ in timing and absolute ∆14C values (Figure 2). Differences in the timing of the ∆14C increases may be due to uncertainties in the age model of the CTFN record of ±1–2 yr. The 20–25‰ differences in absolute ∆14C values in the two Fanning cores may be because the leeward side of the island, where the CTFN core was collected, may have a shallower mixed layer and experience more upwelling than the windward location, where FI5A was collected. Alternatively, the difference in the absolute ∆14C may indicate that there are larger uncertainties in coral ∆14C records, at least at some sites, than had originally been believed (Druffel et al. 1989) and demonstrates the need to ana- lyze ∆14C in multiple corals at some sites. Nonetheless, similar changes in the ∆14C values of each record are consistent with the premise that a common mechanism is controlling the ∆14C records within each coral record. 94 A G Grottoli et al.

Figure 2 Annual coral skeletal ∆14C in FI5A Fanning Island (solid black line) and CTFN Fanning Island (solid dark gray line) (Druffel 1987). * Indicates a single bulk sample for the entire year for FI5A in 1959.

DISCUSSION The positive shift in ∆14C at Fanning Island beginning in 1947 indicates that a shift in the ∆14C composition of the surface waters began at that time. Two potential sources for the high ∆14C to the central equatorial Pacific are discussed below. In addition, we suggest that the phase switch in the Pacific Decadal Oscillation (PDO) beginning in 1942 may have been the mechanism that triggered the rise in ∆14C.

Bomb source of 14C Thermonuclear bombs exploded in the stratosphere between 1955 and 1963 and produced bomb 14C, which was incorporated into seawater and then into coral as it grew. Limited testing of ground-level nuclear weapons took place between 1952 and 1954 that was about 6% of the total bomb 14C produced (950 × 1026 atoms). Druffel (1987) suggested that it was possible for a small amount of close- in fallout of locally-produced bomb 14C from early bomb tests to have caused increases in 14C as early as 1952. High ∆14C values can also occur if post-bomb carbonate recrystalizes in pre-1955 coral skeleton. However, normal to high coral Sr/Ca ratios from 1947–1956 indicate that there was no detectable evidence of recrystalization from aragonite to calcite (D Schrag, unpublished data). The large ∆14C shift at Fanning starting in 1947 appears inconsistent with expected ∆14C changes due to bomb sources and, therefore, seems likely to have had an oceanic origin. Shift in 14C record of an equatorial Pacific coral 95

Table 1 Mean pre-bomb reconstructed Pacific Ocean surface seawater ∆14C values (± 1 standard deviation of the mean), as reconstructed from corals and shells, and as estimated from the distribution of post-bomb dissolved inorganic carbon (DIC) ∆14C values. Region or Current ∆14C (‰) ± 1 SD3 Specific location (lat1, long2) (years) (sample size) Reference and Source Material4 Central Equatorial Pacific Fanning (4′N, 159′W) (1949–1955) –60 ± 7 (7) (Druffel 1987); c Fanning (4′N, 159′W) (1922–1946.8) –55 ± 5 (217) This paper; c Fanning (4′N, 159′W) (1947–1956) –38 ± 8 (78) This paper; c

Northwestern Subtropical Pacific (KC)5 Okinawa (26′N, 127′E) (1912–1954) –38 ± 4 (19) (Konishi et al. 1981); c

Northwestern Tropical Pacific South China Sea –51 ± 9 (10) (Southon et al. 2002); c, b, g (3–17′N, 104–120′E) (1863–1945)

North Pacific Gyre Mid-Gyre Region (25′N, 160′W) –40 (1) (Druffel 1985);6 DIC Hawaii (20′N, 156′W) (1893–1952) –48 ± 4 (74) (Druffel et al. 2001); c

Northeastern Tropical Pacific Panama (7′N, 81′W) (1950–1955) –58 ± 5 (4) (Druffel 1987); c

Southeastern Pacific (SEC/PC)7 Galapagos (0′S, 90′W) (1930–1954) –72 ± 5 (21) (Druffel 1981); c

South Pacific Gyre Mid-Gyre Region (20′S, 160′W) –45 (1) (Druffel 1985);6 DIC Rarotonga (21′S, 160′W) (1950–1956) –52 ± 5 (124) (Guilderson et al. 2000); c Fiji (18′S, 179′E) (1930–1955) –61 ± 7 (14) (Toggweiler et al. 1991); c Fiji (18′S, 179′E) (1950–1955) –62 ± 8 (4) (Toggweiler 1983); c

Southwestern Pacific East Australia (22′S, 153′E) (1950–1955) –54 ± 2 (5) (Druffel and Griffin 1993); c East Australia (23′S, 153′E) (1950–1956) –49 ± 5 (6) (Druffel and Griffin 1993); c 1lat = latitude 2long = longitude 3SD = standard deviation of the mean 4c = coral, b = bivalve, g = gastropod 5KC = Kuroshio Current 6estimates of pre-bomb seawater DIC ∆14C values, as reported by Druffel 1985 7SEC = South Equatorial Current, PC = Peru Current

North Pacific Gyre source of high 14C Estimates of pre-bomb (pre-1955) ∆14C values (± 1 standard deviation) for surface seawater DIC in the Pacific Ocean are listed in Table 1. The lowest value of –72‰ was found at the Galapagos Islands in the eastern South Equatorial Current (SEC) (Druffel 1981) and the highest values of –38‰ were found in the Kuroshio Current (Konishi et al. 1981). Elsewhere in the western Pacific and within the subtropical gyres, ∆14C values ranged between –40‰ and –65‰. In our FI5A core, pre- 96 A G Grottoli et al.

1947 ∆14C values averaged –55 ± 5‰, falling between the estimated ∆14C values in the SEC and the estimated ∆14C values in the mid-northern and mid-southern gyres. After 1947, FI5A ∆14C rose to an average of –38 ± 8‰, approximately matching values in the North Pacific Gyre (NPG) surface waters of –40‰ and the Kuroshio current of –38‰. High ∆14C NPG surface water is subducted and transported southwestward towards the western equatorial Pacific, where it becomes one of the major sources both for the Kuroshio Current and the eastward flowing North Equatorial Counter Current (Liu et al. 1994; Gu and Philander 1997; Zhang et al. 1998; Cai and Whetton 2001; Zhang et al. 2001). Therefore, an increase (decrease) in the pro- portion of NPG waters contributing to equatorial surface waters through vertical advective processes could account for an increase (decrease) in the surface ocean ∆14C at Fanning. However, other recent modeling studies indicate that seawater temperature anomalies arriving at the Equator from the subtropics via this mechanism are likely to be relatively small (Schneider et al. 1999; Pierce et al. 2000; Hazeleger et al. 2001). Variation in the lateral advection of water could also have contributed to the shift in seawater ∆14C at Fanning. Fanning is situated at the boundary between the eastward flowing North Equatorial Counter Current (NECC) and the westward flowing SEC. Expendable bathythermograph (XBT) observations from the 1970s and 1980s suggest that Fanning may be supplied by either the NECC or the SEC depending on the season and on interannual variations in the equatorial Pacific flow (Taft and Kessler 1991). Modeling results suggest that, although seasonal variability in ∆14C in the central equatorial Pacific is relatively small, interannual variability could potentially be large due to the advection of high ∆14C water from further west (Rodgers et al. 1997). High ∆14C NPG surface waters contribute to the more southerly NEC waters and further become entrained in the NECC in the western equatorial Pacific. In addition, relatively low ∆14C water in the portion of the SEC north of the Equator feeds the NECC by turning northward near the date line (Rodgers et al. 1997). A shift in the relative contributions of the SEC and NEC to the NECC could result in a shift in the ∆14C values of surface seawater at Fanning Island. Our Fanning ∆14C data indicate that a significant amount of subtropical seawater is arriving at the Equator, but do not allow us to determine the mechanism for the transport of subtropical seawater to the central equatorial Pacific.

Possible trigger for the shift in 14C Most importantly, the dramatic positive shift in the Fanning ∆14C record beginning in 1947 coincides with the negative shift in the mid-1940s of the Pacific Decadal Oscillation (PDO) (Figure 3) (Mantua et al. 1997). The PDO is an index of slow, large-scale changes in the mode of Pacific Ocean variability with timescales of 20–30 yr, derived from the leading principle component of monthly sea surface temperature (SST) anomalies in the North Pacific Ocean poleward of 20°N (Mantua et al. 1997; Minobe and Mantua 1999). When the PDO is in a warm phase (i.e., the PDO index is positive), the North Pacific has cooler SSTs and higher sea level pressure (SLP) than normal, while in the central and eastern equatorial Pacific, SSTs are warmer and the SLP is lower than normal. The reverse is true during a PDO cool phase. Variations in the PDO correspond to decadal variability generated by ocean-atmosphere processes within the equatorial Pacific (Kirtman and Schopf 1998; Timmermann and Jin 2002). The mid-1940s shift in PDO polarity has been linked to decadal scale changes in a broad range of instrumental and proxy data (Mantua et al. 1997; Zhang et al. 1998; Minobe and Mantua 1999; Biondi et al. 2001; D’Arrigo et al. 2001) but not specifically in the equatorial Pacific. Our Fanning Island ∆14C record suggests that there is a link between the PDO shift in the 1940s and a shift in the water masses arriving in the central equatorial Pacific at that time. Our ∆14C record shifts to a more positive mean value, approximately 4.5 yr after the onset of the negative Shift in 14C record of an equatorial Pacific coral 97

PDO phase (Figure 3). This is consistent with the estimated 3.5–6 yr travel time from the NPG to the Equator reported in modeling studies (Liu et al. 1994; Zhang et al. 2001) and is a bit fast relative to the modeled 6–9 yr travel time calculated by Cai and Whetton (2001).

1.0 -30

-35 ~4.5 yrs 0.5

-40

0.0 -45 C ‰ ( ) 14 ∆

-0.5 -50 PDO Index (5 yrm) ( ) PDO Index (5 yrm)

-55 -1.0

-60 1925 1930 1935 1940 1945 1950 1955 Time (years)

Figure 3 Yearly mean ∆14C and PDO Index from 1922–1956. The PDO Index has been smoothed with a 5-yr (60-month) running mean. Time from the onset of the negative phase in the PDO Index to the increase in yearly mean ∆14C above the long-term pre-1947 mean is approximately 4.5 yr and is indicated by the shaded bar. Dashed lines correspond to the overall mean coral ∆14C from 1922–1947 and 1947–1955. PDO Index data was obtained with permission from the University of Washington’s Joint Institute for the Study of the Atmosphere and Oceans

Synthesis Of the few pre-bomb time series of ∆14C available from the tropical Pacific, most show no major changes associated with the PDO. For example, no sharp transition in ∆14C occurs in the mid-1940s at Hawaii (Druffel et al. 2001), Australia (Druffel and Griffin 1993), Nauru (Guilderson and Schrag 1998b), or the Galapagos (Druffel 1981). However, in the south Pacific gyre, the 140-yr δ18O record from a Moorea (17°30′S, 149°50′W) coral shows a strong interdecadal signal with warmer and/or less saline water in the SPG between 1940 and 1960 (Boiseau et al. 1999). A ∆14C record from this site is not available. In a Rarotonga coral (21°4′S, 159°49′W), the Sr/Ca-derived sea surface temperature record indicates a strong interdecadal component that is synchronous with the PDO Index (Linsley et al. 2000). The limited 1950–55 ∆14C record from the same Rarotonga coral is too short to discern a clear signal prior to 1955 (Guilderson et al. 2000). In the available ∆14C coral records, 98 A G Grottoli et al.

only Fanning indicates a mid-1940s shift. In other available proxy records, only Fanning, Raro- tonga, and Moorea indicate a mid-1940s shift in Sr/Ca or δ18O (Boiseau et al. 1999; Linsley et al. 2000; Dunbar, unpublished). These findings suggest that decadal scale climate fluctuations may be more easily detected in the central Pacific than at its periphery. Overall, near-monthly isotopic measurements in a Porites sp. coral skeleton from Fanning Island revealed a positive shift in the mean ∆14C of 17‰ beginning in 1947 and lasting through at least 1956. The discussion presents intriguing evidence that the positive shift in the ∆14C values of the Fanning Island coral during 1947–1956 is consistent with a decadal timescale introduction of high ∆14C NPG water in the central equatorial Pacific and/or a shift in the proportionate contribution of SEC and NEC surface waters to the NECC. These results suggest that the distinct switch to a negative PDO phase in the mid-1940s could act as the trigger for a shift in ocean circulation. Pre-1950 ∆14C measurements from additional central equatorial corals would enhance our ability to confirm the presence of, and to fully resolve, the extent of the mid-1940s circulation shift in the central Pacific.

ACKNOWLEDGEMENTS Thanks to J Southon, B Frantz, P Zermeno, T Guilderson, and M Kashgarian at CAMS for ∆14C measurements; J Hayes, A McNichol, A Gagnon, and colleagues at NOSAMS for ∆14C measurements; D Schrag for access to his Sr/Ca analyses and advice; K Rogers for his constructive comments on the manuscript; B Linsley for collecting the FI5 core with R Dunbar. This work was funded by a NOAA Office of Global Programs Paleoclimate Program grants, number NA96GP0395 (to ED, SG, AG), a NOAA grant (to RD and B Linsley), an NSF grant OCE9896157 (to RD), the Dreyfus Postdoctoral Fellowship in Environmental Chemistry (to R Cicerone, ED, AG), and the Mellon Foundation.

REFERENCES Barnes DJ, Lough JM. 1993. On the nature and causes of Druffel ERM. 1987. Bomb radiocarbon in the Pacific: density banding in massive coral skeletons. Journal of annual and seasonal timescale variations. Journal of Experimental Marine Biology and Ecology 167:91– Marine Research 45:667–98. 108. Druffel ERM. 1997. Geochemistry of corals: proxies of Biondi F, Gershunov A, Cayan DR. 2001. North Pacific past ocean chemistry, ocean circulation, and climate. decadal variability since 1661. Journal of Climate 14: Proceedings of the National Academy of Science USA 5–10. 94:8354–61. Boiseau M, Ghil M, Juillet-Leclerc A. 1999. Climatic Druffel ERM, Dunbar RB, Wellington GM, Minnis SA. trends and interdecadal variability from South-Central 1989. Reef-building corals and identification of Pacific coral records. Geophysical Research Letters ENSO warming episodes. In: Glynn PW, editor. Vol- 26:2881–84. ume 52. Global Ecological Consequences of the Cai W, Whetton PH. 2001. A time-varying greenhouse 1982–83 El Niño—Southern Oscillation. New York: warming pattern and the tropical-extratropical circula- Elsevier Oceanography Series. p 233–53. tion linkage in the Pacific Ocean. Journal of Climate Druffel ERM, Griffin S. 1993. Large variations of sur- 14:3337–55. face ocean radiocarbon: evidence of circulation D’Arrigo R, Villalba R, Wiles G. 2001. Tree-ring esti- changes in the southwestern Pacific. Journal of Geo- mates of Pacific decadal climate variability. Climate physical Research 98:20,249–59. Dynamics 18:219–24. Druffel ERM, Griffin S, Guilderson TP, Kashgarian M, Druffel ERM. 1981. Radiocarbon in annual coral rings Southon J, Schrag DP. 2001. Changes of subtropical from the eastern tropical Pacific Ocean. Geophysical North Pacific radiocarbon and correlation with cli- Research Letters 8:59–62. mate variability. Radiocarbon 43:15–25. Druffel ERM. 1985. Detection of El Niño and decade ti- Gu D, Philander SGH. 1997. Interdecadal climate fluctu- mescale variations of sea surface temperature from ations that depend on exchanges between the tropics banded coral records: implications for the carbon di- and extratropics. Science 275:805–7. oxide cycle. Geophysical Monographs 32:111–22. Guilderson TP, Schrag DP. 1998a. Abrupt shift in subsur- Shift in 14C record of an equatorial Pacific coral 99

face temperatures in the tropical Pacific associated decadal timescales. Journal of Climate 13:1173–94. with changes in El Niño. Science 281:240–3. Rodgers KB, Cane MA, Schrag DP. 1997. Seasonal vari- Guilderson TP, Schrag DP. 1998b. Radiocarbon variabil- ability of sea surface ∆14C in the equatorial Pacific in ity in the western equatorial Pacific inferred from a an ocean circulation model. Journal of Geophysical high-resolution coral record from Nauru Island. Jour- Research 102:18,627–39. nal of Geophysical Research 103:24,641–50. Schneider N, Venzke S, Miller AJ, Pierce DW, Barnett Guilderson TP, Schrag DP, Goddard E, Kashgarian M, TP, Deser C, Latif M. 1999. Pacific thermocline Wellington GM, Linsley BK. 2000. Southwest sub- bridge revisited. Geophysical Research Letters 26: tropical Pacific surface water radiocarbon in a high- 1329–32. resolution coral record. Radiocarbon 42:249–56. Southon J, Kashgarian M, Fontugne M, Metivier B, Yim Hazeleger W, Visbeck M, Cane M, Karspeck A, Naik N. WW-S. 2002. Marine reservoir corrections for the In- 2001. Decadal upper-ocean temperature variability in dian Ocean and Southeast Asia. Radiocarbon 44:167– the tropical Pacific. Journal of Geophysical Research 80. 106:8971–88. Stuiver M, Polach HA. 1977. Discussion reporting of 14C Hudson JH, Shinn EA, Halley RB, Lidz B. 1976. Sclero- data. Radiocarbon 19:355–63. chronology: a tool for interpreting past environments. Taft BA, Kessler WS. 1991. Variations of zonal currents Geology 4:361–4. in the central tropical Pacific during 1970 to 1987: sea Kirtman BP, Schopf PS. 1998. Decadal variability in level and dynamic height measurements. Journal of ENSO predictability and prediction. Journal of Cli- Geophysical Research 96:12,599–618. mate 11:2804–22. Timmermann A, Jin F-F. 2002. A nonlinear mechanism Konishi K, Tanaka T, Sakanoue M. 1981. Secular varia- for decadal El Niño amplitude changes. Geophysical tion of radiocarbon concentrations in seawater: sclero- Research Letters 29:10,1029/2001GL012269. chronological approach. Proceedings of the Fourth In- Toggweiler JR. 1983. A six zone regionalized model for ternational Coral Reef Symposium 1:181–5. bomb radiotracers in CO2 in the upper kilometer of the Linsley BK, Wellington GM, Schrag DP. 2000. Decadal Pacific Ocean. New York: Columbia University. sea surface temperature variability in the subtropical 421 p. south Pacific from 1726 to 1997 AD. Science 290: Toggweiler JR, Dixon K, Broecker WS. 1991. The Peru 1145–8. upwelling and the ventilation of the South-Pacific Liu Z, Philander SGH, Pacanowski RC. 1994. A GCM thermocline. Journal of Geophysical Research– study of tropical-subtropical upper-ocean water ex- Oceans 96:20467–97. change. Journal of Physical Oceanography 24:2606– Vogel JS, Nelson DE, Southon JR. 1987. 14C background 23. levels in an accelerator mass spectrometry system. Ra- Mantua NJ, Hare SR, Zhang Y, Wallace JM, Francis RC. diocarbon 29:323–33. 1997. A Pacific interdecadal climate oscillation with Weber JN, White EW, Weber PH. 1975. Correlation of impacts on salmon production. Bulletin of the Ameri- density banding in reef coral skeletons with environ- can Meteorological Society 78:1069–79. mental parameters: the basis for interpretation of chro- Minobe S, Mantua N. 1999. Interdecadal modulation of nological records preserved in the coralla of corals. interannual atmospheric and oceanic variability over Paleobiology 1:137–49. the North Pacific. Progress in Oceanography 43:163– Zhang R-H, Kagimoto T, Zebiak SE. 2001. Subduction 92. of decadal North Pacific thermal anomalies in an Nozaki Y, Rye DM, Turekian KK, Dodge RE. 1978. A Ocean GCM. Geophysical Research Letters 28:2449– 200-year record of carbon-13 and carbon-14 varia- 52. tions in a Bermuda coral. Geophysical Research Let- Zhang R-H, Rothstein LM, Busalacchi AJ. 1998. Origin ters 5:825–28. of upper-ocean warming and El Niño change on dec- Pierce DW, Barnett TP, Latif M. 2000. Connections be- adal scales in the tropical Pacific Ocean. Nature 391: tween the Pacific Ocean tropics and midlatitudes on 879–83. RADIOCARBON, Vol 45, Nr 1, 2003, p 101–112 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

DATING OF PREHISTORIC BURIAL MOUNDS BY 14C ANALYSIS OF SOIL ORGANIC MATTER FRACTIONS

Søren M Kristiansen1,2 • Kristian Dalsgaard3 • Mads K Holst4 • Bent Aaby5 • Jan Heinemeier6

ABSTRACT. Dating of prehistoric anthropogenic earthworks requires either excavation for archaeological artifacts or macroscopic organic matter suitable for 14C analysis. Yet, the former, in many cases, is undesirable and the latter is difficult to obtain. Here we present a soil science procedure, which has the potential to overcome these problems. It includes careful sampling of buried former soil surfaces, acid-alkali-acid fractionation of soil organic matter (SOM), and subsequent 14C AMS dating. To test the procedure, soil from one of the largest known burial mounds in Scandinavia, Hohøj, and 9 other Danish burial mounds were sampled. The 14C dates from extracted SOM fractions were compared to reference ages obtained by other methods. We show that humic acid fractions in 7 of the 10 mounds had the same age as the reference, or were, at maximum, 280 yr older than the reference ages. The best age estimates were derived from an organic-rich layer from the upper cm of buried soil or sod. Differences among SOM fraction ages probably indicate the reliability of the dating. Hohøj dated to approximately 1400 BC and, thus, was up to 500 yr older than other dated Scandinavian mounds of comparable size. The remaining investigated burial mounds were dated to between 1700 and 1250 BC. We conclude that combined sampling of buried soil surfaces, SOM fractionation, and 14C analysis allows for dating of archaeological earthworks when minimal disturbance is required, or if no macroscopic organic remains are found.

INTRODUCTION As most preserved historic and prehistoric anthropogenic earthworks in Europe are protected by law, a dating procedure involving minimal disturbance is required. Dating of the soil organic matter (SOM) seems attractive, since many anthropogenic constructions contain buried soil surfaces or sods with moderate or high carbon (C) content. However, dating of bulk samples, particle size fractions, and organic matter fractions from soil samples has been carried out with varying degrees of success in the last 20–25 yr (Scharpenseel and Becker-Heidmann 1992; Wang et al. 1996). Also, chemical fractionation had not been satisfactory in separating labile and refractory soil organic carbon (SOC) pools (Wang and Hseih 2002). However, one procedure has apparently been able to date the burial time of former soil surfaces with some success (Matthews 1980; van Mourik et al. 1999; Dalsgaard and Odgaard 2001). It involves chemical fractionation (acid-alkali-acid extraction) of the SOM prior to 14C analysis and has, so far, yielded the best results in acid, aerobic, sandy soils. The potential of yielding a reliable dating of soil depends on the turnover time of the SOC, which again is a function of the soil type, depth, acidity, land use, and redox potential. Present knowledge unanimously shows that 14C dating of bulk SOM from different depths below the surface does not give an exact numerical age because SOM consists of a continuum of organic materials in all stages of decomposition (Scharpenseel and Becker-Heidmann 1992). 14C dating of organic-rich surface layers (humus layers) containing high contents of C, which originates from plant material with a high decomposition rate, will give a close estimate of the burial time, whereas 14C dating of A-horizon or subsoil overestimates the time of burial, as it merely relies on the SOC’s protection against decomposition. In an archaeological context, this means that the measured 14C date of the time of burial will appear too old. In archaeological excavations, buried humus layers or plant fragments are

1Department of Agroecology, Danish Institute of Agricultural Sciences, P.O. Box 50, DK-8830 Tjele, Denmark. 2Corresponding author. Email: [email protected]. 3Department of Earth Sciences, University of Aarhus, Ny Munkegade Building 520, DK-8000 Aarhus C, Denmark. 4Department of Prehistoric Archaeology, University of Aarhus, Moesgård, DK-8270 Højbjerg, Denmark. 5National Museum Environmental Archaeology, National Museum of Denmark, Ny Vestergade 11, DK-1471 Copenhagen K, Denmark. 6The AMS 14C Dating Laboratory, Institute of Physics and Astronomy, University of Aarhus, Ny Munkegade Building 520, DK-8000 Aarhus C, Denmark.

101 102 S M Kristiansen et al. rare compared to well-preserved A-horizons or organic-containing surface layers and a method suitable for dating of archaeological earthworks should preferably be able to rely on such horizons alone. The approximately 100,000 prehistoric burial mounds recorded in south Scandinavia typically con- sist of numerous sods turned upside-down, and were often constructed on top of an intact soil surface. In the other parts of northern Europe, the number of burial mounds with sods or buried surfaces is probably high, too. Dating the time of construction of prehistoric burial mounds has, in most cases hitherto, been based on archaeological findings and macroscopic organic remains and absolute dating of the construction time of mounds is rare. Furthermore, from previous reviews of soil science in archaeology, it has been concluded that 14C dating of SOC—if it has been mentioned at all—is less useful (Scrudder et al. 1996; Hedges 1994). This conclusion was most probably associated with the problems of selecting an extraction procedure and the estimation of an SOC mean residence time, which makes interpretation difficult. We hypothesize that the time of burial of a mound can be approximated by 14C analysis of the different components of SOM in a buried soil surface after suitable chemical fractionation, since the SOC was probably formed via photosynthesis from atmospheric carbon dioxide at, or shortly before, the time of burial. Hence, the aims of the present study were: i) to test the usefulness of a chemical fractionation of SOM in an archaeological context, and ii) to estimate the time of construction of 10 selected Danish burial mounds with minimal disturbance.

MATERIALS AND METHODS

Field sites and sampling Two separate investigations were performed to test the reliability of 14C dating of the SOM extrac- tions, namely soil samples from a profile in 1 excavated burial mound and samples from auguring in the cores of 9 other mounds (see Table 1). Soil samples were collected from an excavation in the burial mound (named Hohøj) situated in Northern Jutland, Denmark. Like virtually all Danish Bronze Age burial mounds, Hohøj was built of sods only. It has a base diameter of 71 m, a height of 11.7 m, and a volume of approximately 16,630 m3. It is one of the largest known burial mounds in northern Europe. A 10-m deep and 3-m wide excavation was opened in the western side of the burial mound to allow sampling from profiles of undisturbed recent soil processes. Samples from profile walls were collected in November 1998. Bulk soil was collected in 10 × 10 × 30 cm steel frame boxes from 2 places (Hohøj 1 and 2), 400– 450 cm beneath the mound surface, respectively. Both were transported to the laboratory within 24 hr and stored at –18 °C until fractionation. Samples for fractionation were taken from the upper 0.5 to 2 cm of the surface of 2 individual sods in these boxes. The sods in the burial mound had pH (1: 1, soil:water) values of 4.4 to 4.6, and consisted of sand with 5–15% gravel and less than 5% clay. The redox status of the mound was not investigated, but redoximorphic features of Fe- and Mn- oxides were observed in a few, minor spots. Further details on the sampling and the site are given in Aaby and Andreasen (1999) and Bech (2003). As no primary burials were found in the excavation at Hohøj, no reliable archaeological dating of the construction time was possible. Two wood-sticks, identified as heather (Calluna vulgaris), were found in the 14C-dated soil samples after wet sieving. These sticks were considered reliable controls for the 14C dating of SOM, as the maximum time for visible plant remains to decompose is no more than a few decades, and as the C in sticks only resides a few decades in the living biomass before being released to the soil. Dating burial mounds by 14C analysis of soil organic carbon 103

The auguring samples from barrows were obtained from 2 different groups of barrows in the southern part of Jutland, at the parishes of Tobøl and Lejrskov. In both groups, 25 to 30 barrows have been recorded with sizes varying from 10 m in diameter and 1 m in height up to 40 m in diameter and nearly 8 m in height. Today, only 9 barrows are protected by law in each group, while the remaining monuments are under cultivation and quite damaged. All barrows in the 2 groups were surveyed by auguring using a manual chamber auger with a diameter of 7 cm. In the chamber, the individual sods of the barrow construction could be distinguished as systematic sequences of layers. Sometimes, the former vegetation layer could be identified as up to 1-cm-thick, dark, organic-rich layers on top of the individual sods. Disturbances in the barrows appeared as sequences of heterogeneous fill (Bre- uning-Madsen and Holst, forthcoming). In the summer of 2000, samples were consistently taken from cores near the base of the barrows. Preferably, samples were from identifiable former vegetation layers. Otherwise, it was from the topmost 2 cm of a sod or the buried soil surface underneath the barrow. Subsequently, samples from 9 barrows were selected for SOM dating based on the avail- ability of reference dating of the barrows and an evaluation of the quality of the samples, involving criteria such as the extent of disturbances in the barrow, the distinctness of the sample layer, and the certainty of the interpretation of the context of the sample. Samples were stored at –18 °C until fractionation. The dated barrows were all constructed of sandy sods with 2–8% clay and with a pH ranging from 3.4 to 4.6. The barrow Sortehøj consisted of podzolized sods with 1-cm-thick, black, humus horizons on the surfaces. In 1896, during a partial excavation of the barrow, a well-preserved oak-log coffin with grave goods from the Early Bronze Age period III (1300–1100 BC) was uncovered. The sample for the SOM dating was taken from the humus layer of a sod near the base of the barrow. The other barrows consisted of unleached or weakly-leached sods and, consequently, the sod structure was less well-defined than in Sortehøj. Strong redox features in the cores characterized the barrows Skelhøj, Sortehøj, and Lejrskov 2 and 8, where cemented iron pans had formed around an anaerobic environment in the center of the mound. In Skelhøj and Lejrskov 2, the anaerobic environment had preserved plant remains, which were used for reference dating. In the remaining 7 barrows, fragments of charcoal were gathered for maximum age (terminus post quem) reference dating. The charcoal was determined according to the species-level by Claus Malmros, The National Museum of Denmark.

Chemical SOM fractionation In the organic matter fractionation procedure applied here (Dalsgaard and Odgaard 2001), the SOM was separated into 4 compartments: acid-extractable, humic acid, fulvic acid, and residual (humin). Briefly, 4–5 g of soil (with >0.1% C) was acid-washed in 50 ml 0.5% HCl, which yields the acid- extractable organic fraction (Figure 1). The soil was then treated twice with 0.5 M NaOH heated at 80 °C for 2 hr, which left a non-soluble organic fraction, the residual (humin) fraction, and a soluble fraction. Addition of 12 M HCl to the soluble fraction precipitated the humic acid fraction, whereas the supernatant contained the fulvic acid fraction. Precipitated fractions were centrifuged and all fractions freeze-dried prior to 14C dating. Degassed water was used throughout and care was taken during laboratory preparation to avoid contamination from recent atmospheric CO2. However, contaminated, modern C was probably incorporated in the humin fraction due to the high pH in this step (Hatte et al. 2001). Dating of all 4 fractions was only done in 3 samples, whereas humic and fulvic acids fractions were dated in all 10 mounds, as these fractions previously had yielded the best age estimates of burial times (Dalsgaard and Odgaard 2001). The remaining sites, where the residual fraction (which contained >90% of total SOC) was not present, represent SOM pools that are not 104 S M Kristiansen et al. homogenous. Yet, results from Dalsgaard and Odgaard (2001) showed using similar sandy soils, that the humic, fulvic, and residual fractions were virtually even-aged. Table 1 Site and sampling information from the investigated burial mounds in Denmark Sampling depth Site Sb. numbera Location Sample description (m) Hohøj 1 119 56°38′N; 10°00′E Surface of sod 4.20 surface of sodb 4.20 Hohøj 2 119 56°38′N; 10°00′E Surface of sod 4.50 Surface of sodb 4.50 Lejrskov 2 31 55°30′N; 9°18′E Core of burial mound 5.10 Humus layer on sodb 5.35 Lejrskov 8 2 55°31′N; 9°20′E Buried soil surface 3.90 Sodb 3.00 Skelhøj 95 55°25′N; 8°52′E Core of burial mound 4.70 Humus layer on sodb 4.00 Rishøj 61 55°25′N; 8°52′ESod 3.75 Buried soilb 4.00 Jernved 1 56 55°25′N; 8°52′E Buried soil 3.00 Sodb 2.50 Jernved 2 57 55°25′N; 8°52′E Buried soil surface 2.50 Buried soil surfaceb 1.80 Sortehøj 64 55°25′N; 8°51′E Humus layer on sod 1.65 Oak coffinb 3.00 Frishøj 50 55°25′N; 8°51′E Humus layer on soil surface 3.00 Sodb 3.50 Plovshøj 49 55°25′N; 8°51′E Humus layer on sod 2.05 Sodb 2.00 asb. numbers refer to a system of numbering for all Danish burial mounds breference sampling positions

Figure 1 Chemical separation procedure for the soil organic matter fractionation. The obtained fractions are shown by boxes. Dating burial mounds by 14C analysis of soil organic carbon 105

Soil and C analysis The heather sticks and organic C fractions were 14C dated and analyzed for total C content at the AMS 14C Dating Laboratory of the University of Aarhus. Calibrated ages in calendar yr were obtained from 14C ages using the Seattle Program Version 3.03 (Stuiver et al. 1998). Reporting of 14C data was according to Stuiver and Polach (1977). 14C dating of soils are expressed by an apparent mean residence time of the SOM, which is defined as 1/k, where k is the first-order decay constant for organic matter decomposition (e.g., Scharpenseel and Becker-Heidmann 1992).

Calculations The usefulness of C fractionation can be evaluated by comparing the reference dating with the 14C age one would obtain if the soil sample (bulk SOC) was dated without fractionation. The age of the bulk SOC can be calculated from the ages of the fractions by forming a weighted average of the 14C activities, weighted with the masses of the carbon contents. The conversion of age to 14C activity and back follows the expression given by Stuiver and Polach (1977).

RESULTS The conventional 14C ages of the reference samples (heather sticks) from the profile walls in Hohøj were 3180 ± 90 (±1 standard deviation) and 3085 ± 40 BP (Table 2). Heather sticks from the samples obtained by auguring revealed that Lejrskov 2 was erected around 3148 ± 37 BP and Skelhøj around 3185 ± 35 BP. The remaining mounds were reference-dated by charcoal, which was the only available macroscopic organic material. However, charcoal is known to resist microbial decomposition and its C has considerably higher apparent mean residence times compared to bulk SOC. In Lejr- skov 8, Jernved 1, Frishøj, and Plovshøj, pieces of oak (Quercus sp.) dated to 3263 ± 46 BP, 4597 ± 46 BP, 3862 ± 49 BP, and 3391 ± 47 BP, respectively. In Rishøj, charcoal derived from alder (Alnus sp.) dated to 3743 ± 49 BP, and in Jernved 2, a piece from heather dated to 3415 ± 50 BP. Dating of all fractions was only done in Hohøj 1, Lejrskov 2, and Skelhøj. The acid-extractable fraction from Hohøj had the same age (within 1 standard deviation) or was 150 yr younger than the other dated fractions. In Lejrskov 2 and Skelhøj, the acid-extractable fraction was up to 1000 yr older (Table 2). The residual fraction in Hohøj had the same age as the other fractions, and in Lejrskov 2 and Skelhøj, it was 260 and 340 yr older, respectively. Rishøj, Jernved 1, and Frishøj did not follow the same pattern between humic and fulvic acid fractions and reference ages as the other mounds, since they had differences from 700–1520 yr. Differ- ences between fulvic and humic acid fractions in the other 7 mounds were in contrast 460 yr or less. When excluding Rishøj, Jernved 1, and Frishøj, the fulvic acid fractions were within 1 standard deviation (n = 4), or were up to 460 yr older (n = 4) of the reference age. Again, when excluding the 3 mounds mentioned above, the humic acid fractions in the 7 other mounds were within 1 standard deviation of (n = 4), or up to 280 yr older (n = 4) than the reference ages. If excluding samples where reference ages were obtained by charcoal (n = 5), the fulvic and humic acid fractions were within 1 standard deviation, or were 460 and 240 yr older, respectively.

DISCUSSION

The SOM fractionation procedure In 3 mounds—Rishøj, Jernved 1, and Frishøj—there were clearly large discrepancies (>700 yr) between fulvic, humic, and references dates (Figure 2). This suggests that combined SOM fraction 106 S M Kristiansen et al. b parentheses corresponds Calibrated age (cal BC) n. burial mounds in Denmark curve. The age range in 80454548 2200–2150 (2295–2040) 1726–1689 (1742–1637) 1737–1693 (1766–1644) 1681–1640 (1735–1535) 75404743 2880 (2900–2680) 35 1879–1829 (1890–1770) 2026–1981 (2126–1941) 1737–1693 (1766–1664) 1438 (1504–1413) 376039 1425–1414 (1485–1399) 2135–2065 (2200–1985) 1883–1835 (1936–1779) 46 1521 (1603–1462) 45 40 40 45 (1256–1044) 1186–1128 90 1375–1320 (1406–1264) 1381–1321 (1408–1265) 50 1413 (1487–1324) 40 1440 (1520–1320) 1429 (1494–1399) 1386–1322 (1409–1310) ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a C age 14 3770 4215 (BP) Conventional nty of 1 standard deviatio with the calibration organic matter fractions from organic c c C 13 (‰ VPDB) δ , with a measuring uncertai e range of intercepts stick –28.18 3148 stick –28.78 3185 stick –27.4 stick –27.6 3180 3085 sp. charcoal –24.14 3263 yr BP (before present = 1950) Calluna vulgaris Calluna vulgaris Calluna vulgaris Calluna vulgaris Organic matter Organic fraction Quercus Quercus C age. 14 et al. 1998). The first numbers represent th AMS lab number (AAR-) ation in the conventional C dating results for organic remains and extracted soil C dating results for organic 5.105.105.10 73285.5 7329 7330 Residual 7345 Fulvic acid Humic acid4.704.70 –26.84 –26.314.70 –27.36 73324.00 7333 7334 Residual 7346 3401 3421 Fulvic acid 3362 Humic acid –26.98 –26.41 –27.29 3521 3645 3422 4.204.204.20 44304.20 4433 4432 Residual 4288 4.50 Fulvic acid Humic acid 4289 –26.9 –26.9 –26.9 3075 3080 3145 Sampling depth (m) 3.903.00 7344 7553 Humic acid –26.41 3545 14 C ages are given in conventional radiocarbon 14 Hohøj 2 4.50Lejrskov 2 5.10 4434 7327 Humic acid Acid extractable –26.9 –27 3160 Hohøj 1 4.20 4431 Acid extractable –24.3 2930 Locality / Sample location Lejrskov 8 3.90 7343 Fulvic acid –26.24 3715 Skelhøj 4.70 7331 Acid extractable –27 to the ±1 standard devi Calibrated ages in calendar yr (Stuiver Calibrated ages in calendar yr (Stuiver The value assumed Standard Table 2 AMS a b c Dating burial mounds by 14C analysis of soil organic carbon 107 (Continued) d b parentheses corresponds to Calibrated age (cal BC) n. rial mounds in Denmark 120 (1740–1430) 1600–1530 110 (1520–1320) 1430 80 1380–1320 (1430–1220) 55 395 (405–265) 49 2285–2213 (2397–2153) 4939 2186–2141 (2266–2040) 1680–1638 (1688–1540) 494842 2326–2307 (2458–2207) (2138–1977) 2110–2035 2033 (2136–1977) 465065 3362 (3492–3348) 1879–1829 (1917–1750) 1725–1690 (1765–1620) 47 1687 (1741–1623) 504343 1736–1692 (1766–1640) (1290–1128) 1256–1133 1288–1262 (1372–1133) curve. The age range in ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a C age 3290 3170 14 3080 (BP) Conventional rtainty of 1 standard deviatio with the calibration c c c C ganic matter fractions from bu 13 δ (‰ VPDB) e range of intercepts charcoal –26.45 3415 t = 1950), with a measuring unce sp. charcoal –23.53 3391 sp. charcoal –23.59 3862 sp. charcoal –26.81 4597 sp. charcoal –27.38 3743 Organic matter Organic fraction Alnus Quercus Quercus Quercus Quercus Calluna vulgaris C age. 14 radiocarbon yr BP (before presen et al. 1998). The first numbers represent th AMS lab number (AAR-) tion in the conventional C dating results for organic remains and extracted soil or C dating results for organic Sampling depth (m) 4.0 7554 3.50 7555 2.50 7556 3.75 7324 acid Humic –27 3.00 7338 acid Humic –27 3.00 7340 acid Humic –27 2.052.00 7342 7558 acid Humic –27.06 3669 2.502.50 7336 7557 acid Humic –27.10 3400 1.653.00 7325 — acid Humic coffin Oak –28.13 — 3017 — (1300–1100) 14 C ages are given in conventional 14 Locality / Sample location Rishøj 3.75 7323 acid Fulvic –27.06Frishøj 3.00 2320 7337 acid Fulvic –26.51 3824 Jernved 1 3.00 7339 acid Fulvic –26.32Plovshøj 2.05 3356 7341 acid Fulvic –26.35 3678 Jernved 2 2.50 7335 acid Fulvic –26.69 3521 Sortehøj 1.65 7326 acid Fulvic –28.12 2976 the ±1 standard devia Calibrated ages in calendar yr (Stuiver Calibrated ages in calendar yr (Stuiver goods Dated by grave The value assumed Standard Table 2 AMS a b c d 108 S M Kristiansen et al. and 14C analysis yielded an estimate of the reliability of the 14C age, as contamination of SOM or reference sample problems were reflected by age discrepancies. The acid-extractable fraction in these acid soils is suggested to extract young, dissolvable SOC (Wang and Hseih 2002). Younger C may have contaminated the acid-extractable fraction from Hohøj, as this fraction was up to 150 yr younger than the other dated fractions (Table 2). The fractions were probably caused by water-soluble C leached from the mound surface after its construction. In contrast, the acid-extractable fractions from Lejrskov 2 and Skelhøj were up to 1000 yr older than the other dated fractions. This signifies a considerable contamination with older SOC, either derived from water-soluble C in the mound or as the SOC were not homogeneous pools before the sods were placed in the mound. The acid extraction elucidated and, to some degree, probably elim- inated, contamination of the remaining fractions originating from water-soluble C. The residual fraction (humin) is a poorly-known, but highly refractory, pool of SOM (Rice 2001) consisting of insoluble organic remains, e.g. pollen. The residual fraction from the 3 dated samples did not differ from the humic and fulvic acid fractions by more than 250 yr. This signifies that the SOM from the sod surfaces were probably derived from an ecosystem where C was rejuvenated with a considerable speed (i.e., as expected when sampling former soil-surface horizons). The fulvic acid fraction is believed to be a broad group of organic acids, which are mobile in the upper soil during soil formation, e.g. podzolization. Accordingly, although both the fulvic and humic acid fractions had good correlation with references ages, the humic acid ages were probably the most reliable, since the mobility of humic acids in soil is low and the humic acid fraction generally gives the most precise dates of burial time for soil surfaces (van Mourik et al. 1999; Dalsgaard and Odgaard 2001). This accordance between humic and fulvic acid fractions and reference ages was probably related to dynamics turnover times of the different pools of organic C in the original Bronze Age soil. Thus, the humic acid age should be corrected for its apparent mean residence time, which presumably was 280 yr or less. The uncertainty on the time of burial, thus, depended on 14C dating accuracy, the organic matter’s age before incorporation in the SOM, and the mean residence time of the SOM. The latter especially varies greatly between soil types, land uses, vegetation types, climates, etc. (e.g., Wang et al. 1996). The uncertainty introduced by these 280 yr increased the total uncertainty of the archaeological dating considerably but the obtained ages still allow burial mounds to be grouped into archaeological periods, as they cover several centuries. If considering the fractionated samples as 1 bulk sample of the whole soil, the usefulness of the SOM fractionation could be evaluated with respect to 14C dating. For example, we can examine the reference dates obtained from Skelhøj (3185 ± 35 BP) with the humic acid fraction being approximately 250 yr older (Table 2). Calculations on an apparent mean residence time of bulk SOC, as one would obtain if the soil sample was dated without fractionation, gave an age estimate of 3477 BP. In Lejr- skov 2, the (calculated) age of bulk SOC (3382 BP) would not have departed (within 1 standard deviation) from the age obtained by the humic acid fraction (3362 ± 48 BP). However, although the humic acid fraction in Skelhøj overestimated the age by 250 yr, a 14C dating without fractionation would give an additional increase of 55 yr in the estimated age. In addition, without fractionation there would be no indication of the uncertainty or contamination in either of the mounds, since apparent mean residence time of bulk SOM yields no information on this. The 14C age without SOM fractionation would (in all our samples) approach the age of the residual fraction, since it contained 1–2% C, but weighed around 100 times more than the other fractions all together; it contained >90% of the SOC. Hence, SOM fractionation seemed important for both dating reliability and age estimates. Dating burial mounds by 14C analysis of soil organic carbon 109

The visibility of the former vegetation layer or soil surface seemed highly important for precise dating. In the Hohøj and Sortehøj mounds, where the upper 1 cm or less of the former vegetation/humus layer was sampled only, ages of humic acids and references were within 1 standard deviation (Tables 1 and 2). Where the surfaces of the former soil or sod were less clear, as in Lejrskov 2 and 8, Skelhøj, and Plovshøj, the age discrepancy between humic acids and references was up to 280 yr. A precise sampling of an organic-rich, former surface layer, thus, yielded the most reliable samples for SOM fractionation. Such surface layers were probably exclusively built-up of plant remains that were not mixed with C from the subsurface SOC, i.e. on a surface that was not ploughed for decades.

Figure 2 AMS 14C dating results for organic remains and extracted soil organic matter fractions from burial mounds in Denmark. Standard deviations are in Table 2. Note that error bars are omitted, as they are smaller, in most cases, than the symbols.

Problems with 14C dating SOM Problems associated with 14C dating of SOM in burial mounds are the following: • Organic matter near the surface can be relocated by physical and chemical processes; • Certain SOM fractions decompose slowly; • SOC, which originates from above and below ground plant parts, can be recycled by soil biota. The sum of these processes will make organic C in buried soil surfaces appear older than the actual time of burial (e.g., Wang et al. 1996). However, this will cause an error of less than a century (at maximum) when sampling the upper 0.5–2 cm of the former soil surface or humus layer from most ecosystems (Bol et al. 1999; Gaudinski et al. 2000). Furthermore, as the buried soil underneath the burial mounds often show evidence of ploughing by a primitive plough (an ard), sampling should 110 S M Kristiansen et al. also be restricted to the uppermost 1–2 cm here because tillage may increase SOM ages by several centuries (Paul et al. 1997). A careful sampling of the few cm previously at the soil surface is accordingly very important, as our good results from the well-defined surface layer in Sortehøj and Hohøj emphasized. Channels and pores formed by macrofauna and roots should also be avoided, if possible, as they may contain younger C. Soil mixing is often highly localized (Grave and Kealhofer 1999), and hence, can be avoided in profiles below a certain depth. This is typically 0.5–1 m in Dan- ish burial mounds. For samples taken nearer to the present-day surface than the minimum depth of 1.6 m, here the problem of younger C and bioturbation becomes increasingly important, as bioturbation, biological production, and C turnover increase considerably above this depth (e.g. Bird et al. 2002; Gaudinski et al. 2000). Physical transport of the SOM is commonly found in many soils and burial mounds as clay-humus illuviation bands that can be confounded with buried surfaces. A careful macroscopic examination, however, revealed that the upper boundary of the illuviation band was abrupt (a few mm), whereas the upper boundary of sods was more gradual. Thus, erroneous sampling can be minimized and ver- ification can be done by a soil-thin section description. With careful and precise sampling, contamination can be restricted to chemical transport of dissolved organic C and SOC present in the soil before burial. The problem of contamination of younger, water-soluble C, or older SOC present in the soil before burial, can be evaluated by the fractionation procedure applied here, since discrepancies larger than a few decades between the residual, fulvic, and humic fractions apparently signify that the SOC originally was not a homogenous pool. A more advanced SOM fractionation can, in theory, circumvent these problems. This could be done in the following ways: i) by 14C dating of aliphatic hydrocarbons in the SOM, which have low mobility in soils and are formed by plant leaves (Huang et al. 1999); ii) physical fractionation by 14C dating of particulate organic matter (250 to 2000 µm), which have very low apparent mean residence times (e.g. Bird et al. 2002); or iii) by a combination of both.

Archaeological results from the burial mounds Pollen analysis from Hohøj revealed a pollen assemblage typical for Danish burial mounds from the Bronze Age (Aaby and Andreasen 1999). This accords with the ages of the 2 heather sticks, the humic acid, fulvic acid, and residual fractions, all of which give dates around 1400 BC (Table 2; Fig- ure 2). Hohøj was accordingly erected in the Early Bronze Age, probably shortly before 1400 BC. When the total uncertainty is considered, this corresponds to the Late Period II. Thus, Hohøj was approximately 500 yr older than burial mounds of comparable size elsewhere in Scandinavia; although remains of a Swedish burial mound with a similar diameter, but unknown original height, also dates to 1500–1300 BC (Randsborg 1993). The 9 burial mounds explored by soil auguring and buried soil surfaces could not be sampled as precisely as in the case of Hohøj, where a soil profile was available. Nevertheless, dating where reference dates and fulvic and humic fractions were within an interval of <280 yr was established for Lejr- skov 2, Skelhøj, Jernved 2, and Sortehøj. Lejrskov 2 was erected about 1414–1425 BC; Skelhøj was erected around 1438 BC; Jernved 2 most probably was constructed from about 1738–1692 BC; and Sortehøj was correspondingly erected around 1288–1262 BC. More uncertainty exists in age estimates of Lejrskov 8 and Plovshøj, as they had high age discrepancies between fulvic and humic fractions. Yet, Lejrskov 8 was probably constructed around 1600–1450 BC and Plovshøj around 1750– 1600 BC, as the reference age and humic acids (–280 yr) signified. The charcoal found in the burial mounds at Jernved 1 and Rishøj was believed to pre-date the burial mounds’ construction, as they were, respectively, 1240–1520 and 450–1400 yr older than the SOM fractions. This may apply to Dating burial mounds by 14C analysis of soil organic carbon 111

Frishøj as well because of the 700 yr discrepancy between humic and fulvic acid fractions. Accord- ingly, Rishøj was probably constructed around 1600–1300 BC, whereas both Jernved 1 and Frishøj were constructed around 1400–1000 BC. This emphasizes that charcoal as a reference age needs to be interpreted cautiously and seen as a terminus post quem age only, but also that the combination with SOM-fraction dating gives an estimate of the reliability of the charcoal age, or vice versa. All dates fall within the Early Bronze Age (1700–1000 BC) and apparently with a concentration in Period II (1500–1300 BC). It is a relatively narrow timespan compared to the 2-millennia-long timespan from the early 3rd millennium BC to the early 1st millennium BC, where barrows were constructed in large numbers in south Scandinavia. This pattern is interesting in relation to the discussion on how the distinct barrow groups came into existence, since prehistoric barrows often appear concentrated in distinct groups across the landscape in Denmark. From previous excavations, it appears that these barrow groups are characterized by mounds with particularly complex constructions and unusually lavish Bronze Age grave goods. These characteristics are strong indications of areas with a special role in the prehistoric society. The problem with respect to archaeological interpretation of 14C dating is especially bioturbation (e.g., by earthworms), which mixes the soil horizons either before or after burial. Our results and general soil science knowledge suggest that the distinctness of the buried soil layer reflects the SOC dynamics in most soils. A visual inspection of the soil morphology in the profile walls will, thus, in most cases, a priori determine the usefulness of a 14C dating. A soil with low macrofauna activity, as are common in acidic environments and cold or dry climates, thus creates the best conditions for 14C dating of SOM fractions.

CONCLUSIONS Our results showed that even auguring by a hand-operated soil augur gives suitable samples from former surface soils for AMS 14C dating of SOC, though the certainty of the sample’s context inev- itably will be higher in excavations. One advantage of this procedure was that the reliability of 14C dating was already indicated in the field by the distinctness of buried organic-containing layers. The disturbance of the present-day soil surface was minimal and sampling did not change the mounds’ redox status, which in some cases have preserved organic grave goods and corpses for millennia. As many of the northern European burial mounds were constructed from numerous layers of sods, or were constructed above a former soil surface, we conclude that 14C dating of SOM fractions has the potential of being a useful tool in archaeological investigations. However, to achieve a method for archaeological dating that can be applied worldwide, a more advanced fractionation of the SOM is probably required.

ACKNOWLEDGEMENTS Help and comments from Henrik Loft Nielsen, Jens G Bech, Henrik Breuning-Madsen, Niels Clem- mensen, Claus K Jensen, Steffen Terp Laursen, Claus Malmros, and Ernst Stidsing are highly acknowledged. This work was supported by grants from The Danish Ministry of Environment and The Danish Research Councils (The Agrarian Landscape program).

REFERENCES Aaby B, Andreasen E. 1999. Pollenanalytiske undersø- Bech JG. 2003. Fra fortidsminder til kulturmiljø – hvad gelser af bronzealderhøjen, Hohøj, ved Mariager. Na- Alstrup Krat og Hohøj gemte. Copenhagen: Kultur- tionalmuseets Naturvidenskabelige Undersøgelser ministeriet. 200 p. In Danish. 27:1–15. In Danish. Bird M, Santrùcková H, Lloyd J, Lawson E. 2002. The 112 S M Kristiansen et al.

isotopic composition of soil organic carbon on a north- Matthews JA. 1980. Some problems and implications of south transect in western Canada. European Journal 14C dates from a podzol buried beneath an end mo- of Soil Science 53:393–403. raine at Hauga-Breen, southern Norway. Geografiska Bol RA, Harkness DD, Huang Y, Howard DM. 1999. The Annalar 63(A):185–208. influence of soil processes on carbon isotopes distri- Paul EA, Follett RF, Leavitt SW, Halvorsen A, Peterson bution and turnover in the British uplands. European GA, Lyon DJ. 1997. Radiocarbon dating determina- Journal of Soil Science 50:41–51. tion of soil organic matter pool size and dynamics. Soil Breuning-Madsen H, Holst MK. Forthcoming. Soil de- Science Society of America Journal 61:1058–68. scription system for burial mounds—development Randsborg K. 1993. Kivik. Archaeology & Iconography. and uses. Danish Journal of Geography 103. Acta Archaeologica 64:1–147. Dalsgaard K, Odgaard BV. 2001. Dating sequence of Rice JA. 2001. Humin. Soil Science 166:848–57. buried horizons of podzols developed in wind-blown Scharpenseel HW, Becker-Heidmann P. 1992. Twenty- sand at Ulfborg, Western Jutland. Quaternary Interna- five years of radiocarbon dating soils: paradigm of err- tional 78:53–60. ing and learning. Radiocarbon 34(3):541–9. Gaudinski JB, Trumbore SE, Davidson DA, Zheng S. Scrudder SJ, Foss JE, Collins ME. 1996. Soil science and 2000. Soil carbon cycling in a temperate forest: radio- archaeology. Advances in Agronomy 57:1–76. carbon-based estimates of residence times, sequestra- Stuiver M, Polach, HA. 1977. Reporting of 14C data. Ra- tion rates and partitioning of fluxes. Biogeochemistry diocarbon 19(3):355–63. 51:33–69. Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, Grave P, Kaelhofer L. 1999. Assessing bioturbation in ar- Hughen KA, Kromer B, McCormac G, van der Plicht chaeological sediments using soil morphology and J, Spurk M. 1998. INTCAL98 Radiocarbon Age Cal- phytolith analysis. Journal of Archaeological Science ibration, 24,000–0 cal BP. Radiocarbon 40(3):1041– 26:1239–48. 84. Hatte C, Morvan J, Noury C, Paterne M. 2001. Is classi- van Mourik JM, Wartenbergh PE, Mook WJ, Streurmann cal acid-alkali-acid treatment responsible for contam- HJ. 1999. Radiocarbon dating of Paleosols in aerolian ination? An alternative proposition. Radiocarbon sands. Meddedlingen Rijks Geologische Dienst 52: 43(2):177–82. 425–40. Hedges REM. 1994. Radiocarbon dating of soils in ar- Wang Y, Amundson R, Trumbore S. 1996. Radiocarbon chaeology. SEESOIL 10:5–11. dating of soil organic matter. Quaternary Research 45: Huang Y, Li B, Bryant C, Bol RA, Eglinton G. 1999. Ra- 282–8. diocarbon dating of aliphatic hydrocarbons: a new ap- Wang Y, Hseih Y-P. 2002. Uncertainties and novel pros- proach for dating passive-fraction carbon in soil hori- pects in the study of the soil carbon dynamics. Chemo- zons. Soil Science Society of America Journal 63: sphere 49:781–804. 1181–7. RADIOCARBON, Vol 45, Nr 1, 2003, p 113–122 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

BALANCED WINDOW METHOD IN 14C LIQUID SCINTILLATION COUNTING

P TheodÛrsson1 • S Ingvarsdottir • G I Gudjonsson Science Institute, University of Iceland, Dunhaga 3, IS-107 ReykjavÌk, Iceland.

ABSTRACT. The authors present a detailed theoretical and experimental study of the liquid scintillation balanced counting method, widely used in radiocarbon dating, using a simple, laboratory-made system. A fixed counting window becomes a balanced window when the high voltage is set where the 14C count rate rises to a maximum. Using a measured 14C pulse height spectrum, we have calculated the lower and upper limits for 11 balanced windows of varying width and their respective counting efficiencies. Furthermore, we have studied: (1) theoretically and experimentally, the counting efficiency for up to a ±15% shift in pulse height from the balanced setting, (2) the change in pulse height due to temperature variations, (3) the long-time stability of the system, and (4) a method that allows a quick determination of the balance voltage for individual samples, using the Compton spectrum of 133Ba. The standard deviation for thirty 24-hr measuring periods for a 14C standard (190 Bq) was within the expected statistical standard error (0.03%).

INTRODUCTION In high-precision radiocarbon dating, the 14C counting efficiency must be highly stable over the weeks or months it takes to count standards and to date samples. When using a liquid scintillation (LS) counter, the most effective way to attain this is to apply the balanced window method, where the high voltage is set at the maximum 14C count rate in the system’s fixed counting window. Pear- son (1979, 1983) used the balanced window method in his pioneering high-precision LS dating, McCormac has studied it in some detail with a 2-photomultiplier tube (PMT) system (1992) and Arslanov (1991) discusses the method in some detail in his book on 14C dating. The present study is an extension of their work, in which we strengthen the basis of this technique.

METHODS We use a laboratory-built, single PMT system (ICEL-C14) specially designed for 14C dating with an emphasis on simplicity. It consists of a detector unit and a compact, dedicated electronic unit connected to a laptop computer, which controls the counting operations and processes the data. A semi-spherical, 3-ml, quartz vial sits on top of a vertical, 30-mm diameter phototube (TheodÛrs- son 2000). The amplifier of the electronic unit is followed by 4 pulse height discriminators (D1– D4) with inputs connected in parallel order. Their state is read by the computer and when activated, 1 count is added in the respective internal pulse summing registers (N1–N4) of the computer. D1 and D3 define the lower and upper limits of the 14C counting window, denoted by D(L) and D(U). The number of pulses in the 14C window is N(L) – N(U). In the following, L (lower) and U (upper) will refer to discriminator voltages, the corresponding counting registers, or to channel numbers of the 14C spectrum. D2 and D4 are set at 0.60D(U) and 1.25D(U), respectively, and are denoted by D(M) and D(H). Their use will be discussed in the following. We have set the D(L) at 0.9492 V and the D(U) at 5.088 V, giving h = 5.36.

RELATIVE GAIN VERSUS HIGH VOLTAGE AND COUNTING EFFICIENCY In our studies, we use a laboratory-prepared 14C standard (190 Bq in 3 ml of benzene, to which we added 16 g/L of butyl-PBD). We first measured, at varying high voltage (HV), the 14C activity with the ICEL-C14 for the determination of the counting efficiency (ε) given by: ε = [N(L) – N(U)] / T / 190 (1)

1Corresponding author. Email: [email protected]. 113 114 P Theodórsson et al. where T is the counting time in seconds and 190 is the activity of our 14C standard (Bq). ε rises to a maximum at 533.8 V, becoming the balance voltage, denoted as HVbal (Figure 1). At this voltage, the fixed window is called the balanced widow. A shift in the pulse height spectrum will then have a minimal effect on the 14C count rate.

Figure 1 Measured 14C counting efficiency versus high voltage. The solid line is a parabolic fit.

Simultaneously with these measurements, we measured the 14C pulse height spectrum with a multichannel analyzer. From each spectrum, we determined (with interpolation) the channel where the number of counts has fallen to 10% of that in the peak channel (Figure 2). There is a linear relationship between this channel number, Ch(10%), and the HV:

Ch(10%) = –1125.1 + 2.402HV (2)

Figure 2 The channel number where the count rate has fallen to 10% of the maximal count rate versus the high voltage Balanced Window in 14C LSC 115 where Ch(10%) is a measure of the gain and depends on the system used. In the following, we use another gain parameter, the relative gain (RG), which is linearly dependent on Ch(10%) and is instrument independent. We set the RG = 1.00 at HVbal, where Ch(10%) has a value of 157.1. Using Equation 2, we get: RG = –7.162 + 0.01529HV (3) This shows that when the high voltage changes by 1.0 V, the RG changes by 0.015. Using a parabolic fit for the measured counting efficiency (Figure 1), we have calculated the relative counting efficiency (εrel) defined by:

εrel(HV) = ε(HV) / ε(HVbal)(4) Figure 3 displays the deviation of this parameter from its maximum value (1.00) as a function of RG. Since the curve is symmetric at about RG = 1.0, only the part above 1.0 is shown. The figure shows that if the RG shifts 3.5% (corresponding to a change of 2.3 V in the high voltage) from the balance point (RG from 1.00–1.035 or 0.966–1.00), the decrease in the count rate will be 0.1%, corresponding to 8.0 14C yr, a shift that can be tolerated in most cases.

Figure 3 Calculated relative deviation from maximal counting efficiency versus relative gain

14C BALANCED WINDOW In order to calculate the lower and upper limits of a balanced window, an experimentally-determined 14C pulse height spectrum is needed, which we measured with our 14C standard for 5.0 days (Figure 4). To explain the principle of a balanced window let us assume that in Figure 4, D(L) is at channel 20 and D(U) is at channel 120. If a pulse height shift occurs where each pulse is attenuated by 5%, the pulses in channel 20 will not be registered by D(L), nor the pulses in channels 121–126 by D(U). If these 2 changes are equal, the number of pulses in the 14C counting window, N(L)–N(U), will not be affected by this shift. A 14C counting window of this type is called a balanced window. We denote the ratio of channel numbers of the upper- and lower-window limit [Ch(U) and Ch(L)] by h: h = Ch(U) / Ch(L) (5) 116 P Theodórsson et al.

For a selected value of h, the general condition of a balanced window can be described by the following equation: F(L) = hF(U) = hF(hL) (6) where F(Channel number) is a continuous function describing the measured 14C spectrum versus the channel number and L and U are the channel numbers at the lower and upper limits of the balanced window. For a given value of h, Equation 6 unambiguously determines the window limits and the counting efficiency. When we calculate the limits of a balanced window, we use a 6th degree approximation for the measured 14C spectrum (Figure 4). Using Equation 6 for 11 values of h, we calculated the corresponding values of L and U (channel numbers, Table 1). The detection efficiency (ε) was then calculated for the balanced windows using the following equation:

U ε = ∫L F(Ch)dCh / 190 (7)

Figure 4 The pulse height spectrum of a 190-Bq 14C standard measured for 5.0 days. The solid line is a 6th degree fit to the measured spectrum. The discriminator settings for h = 5.36 are shown on the x-axis.

The results of these calculations are shown in Table 1. Furthermore, we have calculated the counting efficiency for a varying shift in gain for h = 5.36 (Figure 5). Table 1 Limits of balanced windows and counting efficiencies Counting h D(L) Channel number D(U) Channel number efficiency Ch(L) keV Ch(U) keV 4.0 32.7 130.9 0.622 30.7 105.7 4.5 30.0 135.1 0.658 28.6 108.8 5.0 27.7 138.7 0.688 26.7 111.4 5.5 26.0 142.9 0.713 25.4 114.6 6.0 24.1 144.6 0.734 23.9 115.8 6.5 22.8 148.2 0.752 22.9 118.5 Balanced Window in 14C LSC 117

Table 1 Limits of balanced windows and counting efficiencies Continued Counting h D(L) Channel number D(U) Channel number efficiency Ch(L) keV Ch(U) keV 7.0 21.3 149.3 0.768 21.7 119.3 7.5 20.3 152.5 0.781 20.9 121.6 8.0 19.1 153.1 0.793 20.0 122.1 8.5 18.4 156.0 0.804 19.4 124.1 9.0 17.4 156.3 0.813 18.6 124.4

Figure 5 Calculated counting efficiency for a balanced window corresponding to h = 5.36 for the varying drift in relative gain

ENERGY CALIBRATION In order to transform the calculated upper and lower channel number limits of the balanced windows (Table 1) to the corresponding energies of the 14C beta particles, we have energy calibrated the system. Due to ionization quenching, the size of the amplified pulses is not strictly proportional to the energy deposited by the 14C beta particles in the scintillator. This quenching increases with decreasing beta energy (Horrocks 1979). At HVbal, we have determined the channel number at 8 different beta particle and electron energies. The result is shown in Table 2 and Figure 6, where we also show points based on similar measurements made by Horrocks (1979). This calibration curve is used to determine the energy limits (keV) of the 11 balanced windows (see Table 1).

Table 2 Energy calibration of the LS counting system keV Channel number X-rays of 133Ba 31.0 31.3 Gamma from 119Sn 31.0 32.0 Gamma from 241Am 59.5 65.3 1% of 14C spectrum 127.9 165.5 3% of 14C spectrum 116.6 146.5 10% of 14C spectrum 95.7 118.2 Max. of 14C spectrum 22.8 25.0 End of 14C spectrum 156.0 182.0 118 P Theodórsson et al.

Figure 6 Energy calibration of the LS system at the balance high voltage (HVbal). Circles are our measured points, triangles are measurements made by Horrocks (1979) fitted to our y-scale. The line is a 2nd order fit to the measured points.

SETTING AND CHECKING THE BALANCE POINT

It is very useful to have a quick and reliable method for determining the HVbal for each sample to be counted, so that the eventual light quenching, vial variation, or vial eccentricity on the photomultiplier tube is corrected for, and also for checking an eventual drift from initial pulse gain at the end of a counting period. For this purpose, we use the Compton spectrum of an external 133Ba gamma source (about 2 µCi), which has a dominating gamma line at 356 keV (Compton edge 207 keV). The ratio b, where: b = N(H) / [N(M) – N(H)] (8) is a good measure of the relative number of pulses in the high part of the Compton spectrum and, therefore, a good indicator of a shift in relative gain (Figure 7). N(H) and N(M) are the number of pulses that have activated discriminators D2 and D4. This is, in principle, the same method that Pearson (1983) used for checking the gain for each counting sample, except that he did not use it to correct the high voltage, but only to check that the shift in gain for individual samples was within a given tolerance limit. At the balance point (533.8 V) and with our discriminator voltage setting, b has a value of 0.86. For fixed ratios of the discriminator voltage, b is presumably independent of the system used. Measuring the 133Ba standard for 2 min gives a relative gain accuracy of 0.01 through the value of b, which corresponds to less than a 0.1% shift in the count rate at the balance setting. If b deviates from 0.86 for a sample to be measured, the high voltage is corrected.

PMT GAIN VERSUS TEMPERATURE CHANGES Figure 3 shows that the 14C counting efficiency in a balanced window changes less than 0.1% (corresponding to 8.0 14C yr) for a ±3.5% shift in the pulse height spectrum from the balance point. Balanced Window in 14C LSC 119

Figure 7 Parameter b = N(H) / [N(M) – N(H)] versus relative gain for an external 133Ba gamma source. The solid line is a 2nd order fit.

Using modern LS counting systems, the change in the 14C count rate in a balanced widow due to any likely drift in high voltage, electronic amplification, or discriminator voltages is negligible. In fact, a significant shift in the pulse height is only to be expected from a drift in internal PMT amplification due to a change in temperature. We have, therefore, studied this effect. The PMT was cooled and heated a few °C below and above the mean ambient temperature. The influence of these variations on the internal gain was determined 137 by the shift of the Compton edge of an external Cs gamma source at HVbal using a multichannel analyzer. From the results (Figure 8), we find that the relative internal gain of the PMT decreases linearly by –0.0046 per °C. If we want to keep the decrease in counting efficiency within 0.1% from its balance point value, the temperature should be kept within ±8 °C from the initial temperature, a condition not difficult to fulfill.

Figure 8 Shift in channel number at 137Cs Compton edge versus temperature of the PMT 120 P Theodórsson et al.

LONG-TERM STABILITY AND REPRODUCIBILITY

The long-term stability of the ICEL-C14 was studied by first setting the high voltage at HVbal. Our 14C benzene standard was then measured continuously at 24-hr counting periods for a month (Figure 9). The small, gradual decline in the count rate is due to the loss of benzene through evaporation, about 0.13% over 30 days (0.9 mg/week). The standard deviation from the line of best fit is – 0.023%. As we count 11.2 106 pulses in the 14C window in each 24-hr counting period, the relative statistical standard deviation is 0.030%. Even at this high level of precision, we can see no drift from a constant value. These measurements prove convincingly the extremely high stability of the ICEL-C14 at the balance setting, and is well below what is needed in high-precision dating.

Figure 9 Stability test using consecutive 24-hr measurements of a 14C standard (190 Bq)

The counting data contains information that can be used for checking the drift in relative gain, initially set close to 1.00, during the 30-day testing period. We use a parameter t, defined by: t = [N(U) – N(H)] / [N(L) – N(U)] (9) which measures the relative number of pulses in the tail of the 14C spectrum. We determine the value of t versus the relative gain by measuring the 14C standard at varying high voltages, i.e., by using the same data as for determining the counting efficiency, as described above. There is a linear relationship between t and the relative gain in the interval of 0.85 to 1.15: RG = 0.822 + 3.27t (10)

At HVbal (RG = 1.00), t has a value of 0.056. During the 30-day stability test period, the number of pulses in the 4 counting channels of ICEL-C14 was recorded in 6-hr intervals, from which the value of t was determined (Figure 10). There was a small, gradual increase in the value of t, corresponding to a drift from an initial relative gain of 1.00 to 0.987 at the end of the 30-day period (September), which could have been caused by a 2.8 °C decrease in temperature. This shift in relative gain should, according to Figure 3, cause a decrease of 0.012% in counting efficiency at the end of the period. The 14C count rate reproducibility of sample changing was studied by setting the system initially at the balance point, and then counting the 14C benzene standard 10 times in 23-hr periods, removing Balanced Window in 14C LSC 121

Figure 10 The measured value of parameter t = [N(U) – N(H)] / [N(L) – N(H)] during the 30-day stability test period and reinstalling the sample between measurements (Figure 11). After correction for evaporation, we get: • Mean 14C count rate 130.17 cps • Measured standard deviation 0.040% • Calculated relative statistical deviation 0.030% •Value of t 0.052–0.054 Finally, we measured the sample twice at 23-hr periods, where the vial had an obvious eccentricity of 2–3 mm relative to the axis of the PMT. This displacement reduced the count rate by 0.07% (Fig- ure 11), while the t value was 0.049. These measurements show that no special care is needed in placing the vial on the top of the PMT, even in work at the highest practical precision.

Figure 11 Repeated 23-hr measurements for 14C standard (190 Bq). The vial was removed and reinserted between measurements. Circles: vial centered; triangles: vial displaced 2–3 mm from center 122 P Theodórsson et al.

CONCLUSION When an LS counting system is used in 14C dating, a high counting stability is best secured by apply- ing the balanced window method. The balance point can be determined (or checked) quickly for each sample by using an external gamma source. For our system, a change of ±8 °C from the temperature when the balance voltage is set will decrease the count rate in the 14C window by 0.1%. The count rate drift in the 14C window was less than 0.03% during the 30-day stability test period and the reproducibility in sample changing was better than 0.015%.

REFERENCES Arslanov K. 1991. Radiocarbon: geochemistry and geo- McCormac FG. 1992. Liquid scintillation counter char- chronology. Leningrad: Leningrad University Press. acterization, optimization, and benzene purity correc- In Russian. tion. Radiocarbon 34(1):37–45. Gudjonsson GI, TheodÛrsson P. 2000. A compact auto- Pearson GW. 1979. Precise 14C measurement by liquid matic low-level liquid scintillation system for ra- scintillation counting. Radiocarbon 21(1):1–21. don-in water by pulse pair counting. Applied Radia- Pearson GW. 1983. The development of 14C measurement tion and Isotopes 53:377–80. and its application to archaeological timescale prob- Horrocks DL. 1979. Energy per photoelectron in a coin- lems. PhD dissertation. Belfast: The Queen’s Univer- cidence liquid scintillation counter as a function of sity of Belfast. electron energy. In: Peng C-T, Horrocks DL, Alpen TheodÛrsson P. 2000. A multi-sample liquid scintillation EL, editors. Liquid scintillation counting: recent ap- counting system for weak beta emitting samples with plications and development. New York: Academic single phototube detectors. Applied Radiation and Press. p 16–29. Isotopes 53:297–301. RADIOCARBON, Vol 45, Nr 1, 2003, p 123–130 © 2003 by the Arizona Board of Regents on behalf of the University of Arizona

RADIOCARBON DATES FROM HALFIAH GIBLI (ABADIYEH), A PREDYNASTIC SETTLEMENT IN UPPER EGYPT

Kathryn A Bard Department of Archaeology, Boston University, 675 Commonwealth Avenue, Boston, Massachusetts 02215, USA. Email: [email protected].

ABSTRACT. In 1989 and 1991, wood charcoal samples were excavated at a Predynastic settlement in Upper Egypt, Halfiah Gibli (HG). A second site, Semaineh (SH), was also investigated, but as the ceramics there were mostly from the Old King- dom, excavations were concentrated at HG. Wood charcoal was obtained in undisturbed contexts, in association with Nagada culture potsherds and lithics, ranging in date from about 3700 BC to 3200/3100 BC. These new radiocarbon dates provide more data for the relative phases of the Nagada culture, formulated mainly from ceramic seriation.

INTRODUCTION The Predynastic period in Egypt, when complex society arose, spans most of the 4th millennium BC, culminating in about 3100–3000 BC with the emergence of Egypt’s Early Dynastic state. While a number of Predynastic cemeteries have been excavated, beginning with Flinders Petrie’s excavation of over 2000 burials at Nagada in 1894–95 (Petrie and Quibell 1896), Predynastic settlements have not been well preserved and their excavation was frequently overlooked in favor of the much more spectacular burial evidence. In 1989, initial investigations (surface survey and several 1 × 1 m test excavations) were conducted by K Bard at the Predynastic settlements of Halfiah Gibli (HG) and Semaineh (SH) in Upper Egypt (26°00′N, 32°22′E). These 2 Predynastic sites were later excavated in 1991. Sites HG and SH were first mentioned by Petrie in his well-known volume, Diospolis Parva, The Cemeteries of Abadiyeh and Hu (1901), in which his Sequence Dating system for seriating Predy- nastic pottery was explained. This ceramic seriation system was later modified by Werner Kaiser (1957), and as a result, the 3 relative phases of Predynastic culture are now called Nagada I, II, and III, after the largest-known Predynastic site. More recently, several studies of 14C dates from settlements have been published by Hassan (Hassan 1984; Hassan 1985; Hassan and Robinson 1987), and an absolute chronology of the Nagada culture relative phases has been collated by Hendrickx (1996: 64) based on Hassan’s radiocarbon dates. 14C dates obtained from wood and matting excavated at Abydos (cemeteries U, B, and the Early Dynastic royal cemetery) by the German Institute of Archaeology, Cairo, provide an even longer sequence of calibrated dates, from the Nagada Ib phase to the end of the 1st Dynasty but with a long gap in dates between the Nagada Ic and Nagada IId phases (Görsdorf et al. 1998).

Table 1 Hendrickx’s Absolute Chronology of the Predynastic Nagada Culture Period cal BC Nagada Ia–IIb approximately 3900–3650 BC Nagada IIc–IId2 approximately 3650–3300 BC Nagada IIIa1–IIIb/Dyn. 0 approximately 3300–3100 BC

The purpose of this article is to provide further correlation between the relative dates of Nagada culture phases and 14C dates.

123 124 K Bard

METHODS Situated on spurs of the low desert above the floodplain and to the south of the el-Ranan canal, both HG and SH sites were deflated and had been disturbed by modern cultivation and/or farming activities (Bard 1989). In 1989 and 1991, charcoal samples were collected from deposits in the remaining areas that did not seem to have been disturbed by later human activity. During the excavations, relative dates were given to the excavated deposits using Kaiser’s (1957) seriation of Predynastic Nagada culture pottery classes. The 1991 excavations at SH revealed a site with a great mixture of ceramics, predominantly dating to the Old Kingdom (about 2686–2125 BC) but mixed with a few Predynastic and New Kingdom sherds. As SH seemed to be an Old Kingdom site, excavations were discontinued there. Petrie had excavated a cemetery area (H) at Semaineh on a small spur east of the settlement and this area is probably where the mainly Nagada III grave goods were found (Bard 1992). The 1991 excavations at HG revealed the settlement associated with the Predynastic cemetery (Abadiyeh) excavated by Petrie. There was no evidence of any structures at the site and 9 units were excavated in the few undisturbed deposits, i.e., in margins to the north and east of the main spur (Bard 1992). All of the ceramics excavated at HG are from the Nagada Ic to Nagada IIb–c phases (Sally Swain, personal communication). Excavated charcoal samples were collected with metal trowels, and were then wrapped in aluminum foil and placed in plastic sample bags. The 1989 charcoal samples were submitted to the Oxford Radiocarbon Accelerator, Research Lab- oratory for Archaeology. The 1991 charcoal samples were submitted to the Radiocarbon Laboratory of the Institute for the Study of Earth and Man at Southern Methodist University (Dallas, Texas, USA), under the direction of Herbert Haas. This laboratory later moved and became the WRC Radiocarbon Laboratory, Water Resources Center, Desert Research Institute (Las Vegas, Nevada, USA). Table 2 Calibrated 14C dates from the predynastic sites of Halfiah Gibli and Semaineh Calibration Sample No. Lab No. BP 68.2% 95.4% HG-1 OxA-2182 4590 ± 80 BP 3510 BC (19.8%) 3420 BC 3650 BC (95.4%) 3000 BC 3390 BC (19.5%) 3300 BC 3240 BC (28.9%) 3100 BC HG-2 OxA-2183 4810 ± 80 BP 3700 BC (2.8%) 3680 BC 3760 BC (78.7%) 3490 BC 3670 BC (61.5%) 3510 BC 3470 BC (16.7%) 3370 BC 3400 BC (3.9%) 3380 BC SH-3 OxA-2184 4860 ± 80 BP 3760 BC (46.5%) 3620 BC 3950 BC (95.4%) 3350 BC 3600 BC (21.7%) 3520 BC SH-4 OxA-2185 4020 ± 80 BP 2850 BC (4.7%) 2810 BC 2900 BC (95.4%) 2300 BC 2670 BC (61.8%) 2450 BC 2420 BC (1.7%) 2400 BC HG-50 DRI-2833 4604 ± 91 BP 3520 BC (46.2%) 3310 BC 3650 BC (95.4%) 3000 BC 3240 BC (12.0%) 3170 BC 3160 BC (10.0%) 3100 BC SH-150 DRI-2907 4933 ± 136 BP 3950 BC (17.7%) 3840 BC 4050 BC (95.4%) 3350 BC 3820 BC (47.7%) 3630 BC 3560 BC (2.8%) 3530 BC Dates from Halfiah Gibli (Abadiyeh), Upper Egypt 125

Table 2 Calibrated 14C dates from the predynastic sites of Halfiah Gibli and Semaineh Continued Calibration Sample No. Lab No. BP 68.2% 95.4% HG-332 DRI-2834 5060 ± 110 BP 3970 BC (63.5%) 3750 BC 4250 BC (1.7%) 4100 BC 3740 BC (4.7%) 3710 BC 4050 BC (93.7%) 3600 BC HG-341 ETH-13011 4680 ± 65 BP 3620 BC (5.3%) 3600 BC 3640 BC (95.4%) 3340 BC 3530 BC (62.9%) 3360 BC HG-349 DRI-2835 4290 ± 140 BP 3100 BC (43.5%) 2830 BC 3350 BC (95.4%) 2450 BC 2820 BC (22.5%) 2660 BC 2650 BC (2.2%) 2620 BC HG-526 DRI-2906 4498 ± 131 BP 3370 BC (66.6%) 3010 BC 3550 BC (95.4%) 2850 BC 2980 BC (0.8%) 2970 BC 2950 BC (0.8%) 2930 BC HG-545 SMU-2754 4860 ± 70 BP 3720 BC (46.8%) 3620 BC 3800 BC (92.5%) 3500 BC 3590 BC (21.4%) 3520 BC 3430 BC (2.9%) 3380 BC HG-655 DRI-2905 4731 ± 132 BP 3650 BC (68.2%) 3360 BC 3800 BC (95.4%) 3050 BC

RESULTS, 1989 SAMPLES Four samples of wood charcoal were taken in test pits at HG and SH. The calibrated dates of these samples were previously published (Bard 1991), but have been re-calibrated here using OxCal v3.8 (Bronk Ramsey 2002). HG-1, OxA-2182 14C age: 4590 ± 80 BP Cal age(s), 68.2% probability: 3510 BC (19.8%) 3420 BC 3390 BC (19.5%) 3300 BC 3240 BC (28.9%) 3100 BC 95.4% probability: 3650 BC (95.4%) 3000 BC Wood charcoal taken 5–10 cm below the surface in an ancient midden in the uncultivated area of HG, at the bottom of a small wadi full of Predynastic sherds and lithics, which had washed down the slope. Collected 1989 by K Bard. Comment: Calibrated dates of this sample fall within Hendrickx’s phase of Nagada IIc–IId2, about 3650–3300 BC, the middle Nagada phase. HG-2, OxA-2183 14C age: 4810 ± 80 BP Cal age(s), 68.2% probability: 3700 BC (2.8%) 3680 BC 3670 BC (61.5%) 3510 BC 3400 BC (3.9%) 3380 BC 95.4% probability: 3760 BC (78.7%) 3490 BC 3470 BC (16.7%) 3370 BC Wood charcoal taken 5–10 cm below the surface in a small, ancient midden in the uncultivated area of HG. Collected 1989 by K Bard. Comment: Calibrated dates of this sample fall within Hendrickx’s phase of Nagada IIc–IId2, about 3650–3300 BC, the middle Nagada phase, but earlier than sample HG-1, OxA-2182. SH-3, OxA-2184 14C age: 4860 ± 80 BP Cal age(s), 68.2% probability: 3760 BC (46.5%) 3620 BC 3600 BC (21.7%) 3520 BC 95.4% probability: 3950 BC (95.4%) 3350 BC Wood charcoal taken 5–10 cm below the surface of a test pit in an area south of a mudbrick feature visible on the surface of the site. Collected 1989 by K Bard. 126 K Bard

Comment: This date is from the earliest phase of Predynastic culture (Hendrickx’s Nagada Ia–IIb, about 3900–3600 BC). SH-4, OxA-2185 14C age: 4020 ± 80 BP Cal age(s), 68.2% probability: 2850 BC (4.7%) 2810 BC 2670 BC (61.8%) 2450 BC 2420 BC (1.7%) 2400 BC 95.4% probability: 2900 BC (95.4%) 2300 BC Wood charcoal taken 5–10 cm below the surface in a test pit within a rectangular feature of decayed mudbrick. Collected 1989 by K Bard. Comment: This feature can now be dated to the Old Kingdom. The presence of predominantly Old Kingdom sherds on the surface of this site helps to explain this date.

Figure 1 14C dates for Halfiah Gibli (HG) and Semaineh (SH). Atmospheric data from Stuiver et al (1993), Oxcal v3.8 Bronk Ramsey (2002), cub r:4 sd:12 prob usp[chron]

RESULTS, 1991 SAMPLES Eight samples of wood charcoal were taken from excavated units at HG and SH. A 9th sample (HG- 399, DRI 2836) “came out very small, and its conversion to benzene was a speculative undertaking” (Haas, personal communication). Hence, this sample is not listed here. Calibration uses OxCal v3.8 (Bronk Ramsey 2002). Dates from Halfiah Gibli (Abadiyeh), Upper Egypt 127

HG-50, DRI-2833 14C age: 4604 ± 91 BP Cal age(s), 68.2% probability: 3520 BC (46.2%) 3310 BC 3240 BC (12.0%) 3170 BC 3160 BC (10.0%) 3100 BC 95.4% probability: 3650 BC (95.4%) 3000 BC Wood charcoal excavated in a deposit of gravelly silt about 15–30 cm below the surface that had been washed down from the main settlement along with a number of Predynastic artifacts, mainly potsherds and lithics. Collected 1991 by H Raab-Rust. Comment: This redeposited sample and associated artifacts suggest movement of village artifacts and ecofacts by wadi activity after Nagada culture occupation of the site. SH-150, DRI-2907 14C age: 4933 ± 136 BP Cal age(s), 68.2% probability: 3950 BC (17.7%) 3840 BC 3820 BC (47.7%) 3630 BC 3560 BC (2.8%) 3530 BC 95.4% probability: 4050 BC (95.4%) 3350 BC Wood charcoal sample found next to a Meydum ware sherd (Old Kingdom) 5–10 cm below the surface. Other pottery in this feature, which was thought to be a kiln for Old Kingdom bread molds, was a great mixture of Old and New Kingdom wares with some Predynastic potsherds. The majority of sherds were Old Kingdom bread molds. Collected 1991 by S Savage. Comment: Although this sample was collected in an Old Kingdom context, its calibrated dates are much earlier, suggesting a disturbed feature with a mixture of artifacts from different periods (Pre- dynastic and Old Kingdom). HG-332, DRI-2834 14C age: 5060 ± 110 BP Cal age(s), 68.2% probability: 3970 BC (63.5%) 3750 BC 3740 BC (4.7%) 3710 BC 95.4% probability: 4250 BC (1.7%) 4100 BC 4050 BC (93.7%) 3600 BC Wood charcoal associated with a semi-circular hearth approximately 77 cm below the surface. Col- lected 1991 by H Raab-Rust. Comment: No pottery or lithics were found associated with this hearth, excavated well below the stratigraphic location of all Predynastic artifacts. It is the earliest date obtained at this site, possibly predating Nagada culture occupation there. HG-341, ETH-13011 14C age: 4680 ± 65 BP Cal age(s), 68.2% probability: 3620 BC (5.3%) 3600 BC 3530 BC (62.9%) 3360 BC 95.4% probability: 3640 BC (95.4%) 3340 BC Wood charcoal taken in a test pit approximately 40 cm below the surface (and below the plow zone) in the Predynastic settlement that had been disturbed by cultivation in the 1950s and 1960s. Col- lected 1991 by K Bard. Comment: This sample was taken just above the dark red paleosol, and although the artifacts in the test pit demonstrate post-Predynastic disturbance, the calibrated dates of the sample suggest a Nagada IIc–IId2 date for the occupation of the HG village (see Hendrickx 1996:64). 128 K Bard

HG-349, DRI-2835 14C age: 4290 ± 140 BP Cal age(s), 68.2% probability: 3100 BC (43.5%) 2830 BC 2820 BC (22.5%) 2660 BC 2650 BC (2.2%) 2620 BC 95.4% probability: 3350 BC (95.4%) 2450 BC Wood charcoal taken in a deposit approximately 5–20 cm below the surface in which the only sherds of marl ware were found at HG. Marl ware is more commonly found in Predynastic burials and is much less common in settlement contexts. Collected 1991 by J Raab-Rust. Comment: The deposit contained much ash and few artifacts, which the project paleoethnobotanist, Wilma Wetterstrom, thought might represent ashy deposits that were removed from village houses and dumped on the edge of the village (Wetterstrom, personal communication). The calibrated dates of the sample, however, are later than the Predynastic. HG-526, DRI-2906 14C age: 4498 ± 131 BP Cal age(s), 68.2% probability: 3370 BC (66.6%) 3010 BC 2980 BC (0.8%) 2970 BC 2950 BC (0.8%) 2930 BC 95.4% probability: 3550 BC (95.4%) 2850 BC Wood charcoal excavated about 5–10 cm below the surface in a deposit in the undisturbed area of the main Predynastic settlement, just above the dark red paleosol. Collected 1991 by J Raab-Rust. Comment: The calibrated dates would suggest the latest occupation of the Predynastic village, in the Nagada IIIa1–IIIc1 phases (see Hendrickx 1996:64). HG-545, SMU-2754 14C age: 4860 ± 70 BP Cal age(s), 68.2% probability: 3720 BC (46.8%) 3620 BC 3590 BC (21.4%) 3520 BC 95.4% probability: 3800 BC (92.5%) 3500 BC 3430 BC (2.9%) 3380 BC Wood charcoal excavated about 10–15 cm below the surface in the area of a lithics workshop, in which there was abundant charcoal and lenses of ash. Collected 1991 by K Bard. Comment: Also in this feature was the end fragment of a ground greywacke rhomboid-shaped palette, as found in Predynastic graves dating to late Nagada I and early Nagada II phases (about 3650 BC, Hendrickx 1996:64). The relative dates of these burials would accord well with the calibrated dates of the sample. HG-655, DRI-2905 14C age: 4731 ± 132 BP Cal age(s), 68.2% probability: 3650 BC (68.2%) 3360 BC 95.4% probability: 3800 BC (95.4%) 3050 BC Wood charcoal excavated approximately 10–15 cm below the surface in deposits of overlapping ash and sand lenses with abundant charcoal, lithics, and a Nagada Ic C-class sherd. Collected 1991 by S Savage. Comment: The calibrated age of 3650 BC would be appropriate for the Nagada Ic sherd.

CONCLUSION Wood charcoal taken from the excavated units at site HG is Predynastic in date, whereas wood charcoal from site SH is Predynastic and Old Kingdom in date. The contexts of these samples indicate Dates from Halfiah Gibli (Abadiyeh), Upper Egypt 129 disturbance of the sites in ancient and modern times and at HG, there is also evidence of slope wash and erosion. There is no evidence at HG of any kind of permanent settlement, however, and occupation may have been sporadic and/or seasonal. Given a lack of such evidence, possibly another more permanent settlement existed on the floodplain and is now destroyed or covered with river sediments. The existence of such a site would explain the presence of a large Predynastic cemetery (B) that Petrie excavated at Abadiyeh with a number of high-status burials. A calibrated date of 3800–3500 BC (HG-545, SMU-2754) for wood charcoal from a lithics workshop, in which a fragment of a rhomboid-shaped palette was found, accords well with a relative date of Nagada Ic–IIa for this artifact. This calibrated date, along with that of another sample (HG-655, DRI-2905) associated with a Nagada Ic C-class potsherd, would place the transition from Nagada I to Nagada II at about 3600 BC. Wood charcoal (HG-332, DRI-2834) excavated at HG under a deep deposit of sterile, windblown sand, and associated with a hearth but no potsherds or lithics, provides evidence of the earliest use of the site, possibly predating the occupation of Nagada culture peoples. Both sites HG and SH were in use in the Old Kingdom, as the calibrated dates of several samples demonstrate, at which time there was much disturbance of the earlier Predynastic debris. Calibrated dates of samples from HG of the 4th millennium BC indicate a Predynastic village occupied from about 3700 BC (HG-545, SMU-2754) to about 3200/3100 BC (HG-526, DRI-2906). Although the site was disturbed in ancient and recent times, wood charcoal samples collected in contexts with Predynastic potsherds and lithics provide more relative and absolute dates for the Pre- dynastic Nagada culture.

ACKNOWLEDGMENTS Funding for this project was provided by grants from the National Geographic Society, Committee for Research and Exploration. The author gratefully acknowledges their support of this project. I would also like to thank Andrea Manzo of the Instituto Universitario Orientale, Naples, Italy, for calibrating the dates BC of the 12 charcoal samples from sites HG and SH, which he very gener- ously offered to do in 2000. In 2003, Larry Pavlish of the Department of Physics (Isotope), Univer- sity of Toronto, calibrated the dates of the 12 samples using an updated program. Sally Swain was the project’s ceramic analyst who studied the excavated potsherds and placed them in a relative sequence based on Werner Kaiser’s seriation (1957).

REFERENCES Bard KA. 1989. Predynastic settlement patterns in the search Center in Egypt 158/159:11–5. Hu-Semaineh Region, Egypt. Journal of Field Ar- Hassan FA. 1984. Radiocarbon chronology of Predynas- chaeology 16:475–78. tic Naqada settlements, Egypt. Current Anthropology Bard KA. 1991. Egypt, Halfiah Gibli and Semaineh H, 25:681–3. Hiw. In: Hedges REM, Houseley RA, Bronk RA, van Hassan FA. 1985. Radiocarbon chronology of Neolithic Klinken GJ. Radiocarbon dates from the Oxford AMS and Predynastic sites in Upper Egypt and the Delta. System: Archaeometry Datelist 12. Archaeometry African Archaeological Review 3:95–116. 33(1):129–30. Hassan FA, Robinson SW. 1987. High-precision radio- Bard KA. 1992. Preliminary report: the 1991 Boston carbon chronometry of ancient Egypt and compari- University excavations at Halfiah Gibli and Se- sions with Nubia, Palestine and Mesopotamia. Antiq- maineh, Upper Egypt. Newsletter of the American Re- uity 61:119–35. 130 K Bard

Hendrickx S. 1996. Relative chronology of the Naqada Kaiser W. 1957. Zur inneren Chronologie der Culture. Problems and possibilities. In: Spencer J. As- Naqadakultur. Archaeologia Geographica 5/6:69–77. pects of Early Egypt. London: British Museum Press. Petrie WMF, Quibell JE. 1896. Naqada and Ballas. Lon- p 36–69. don: British School of Archaeology in Egypt. Görsdorf J, Dreyer G, Hartung U. 1998. New 14C dating Petrie WMF. 1901. Diospolis Parva. The Cemeteries of of the Archaic Royal Necropolis Umm el-Qaab at Abadiyeh and Hu. London: Egypt Exploration Fund. Abydos (Egypt). Radiocarbon 40(2):641–7. RADIOCARBON UPDATES

We regret to announce the passing of an important member of the radiocarbon family, John Head. Radiocarbon will publish an obituary in the next issue, Vol 45, Nr 2, 2003.

Next Radiocarbon issue Our next issue is devoted to the much-anticipated results from the Fourth International Radiocarbon Intercomparison (FIRI), guest edited by E Marian Scott.

Closing of Radiocarbon Laboratory at University of California, Riverside After 30 years of operating in the Department of Anthropology and Institute of Geophysics and Planetary Physics at the University of California, Riverside (UCR), and anticipating retirement, Emeritus Professor, R E (Erv) Taylor has terminated the decay counting operations of the UCR Radiocarbon Laboratory and will be moving other components to two other University of California campuses. Instrumentation involved in the preparation of graphitic carbon for AMS-based 14C dating and an HPLC system for biochemical characterization of skeletal materials will become part of the laboratory of David Glen Smith at the University of California, Davis (UCD). Professor Smith will also collaborate in a future project involving mitrochondrial DNA studies of human bone and teeth. Pro- fessor Taylor’s last graduate student, who is currently undertaking her dissertation research, has moved to UCD to continue the operation of this instrumentation and be associated with Professor Smith’s laboratory. One of the UCR graphite processing lines will be moved to the Keck Carbon Cycle Accelerator Mass Spectrometry Laboratory in the Department of Earth System Science at the University of Cal- ifornia, Irvine (UCI) under the direction of Ellen Druffel, John Southon, and Susan Trumbore. It is anticipated that the last UCR Radiocarbon Laboratory supervisor will join the staff of the Keck AMS laboratory at UCI in the summer of 2003. The UCI Keck AMS Laboratory is continuing the seventy year tradition of University of California involvement in the 14C field dating back to the work of Libby and other researchers at University of California, Berkeley in the 1930s and subsequently carried out at the Los Angeles, San Diego, Riverside, and Irvine campuses. Professor Taylor is now associated with the Cotsen Institute of Archaeology (CIOA) at UCLA and will continue several research projects under CIOA auspices. His other responsibilities will include spoiling his grandchildren, waiting for his first great-grandchild, and attempting to complete a long- overdue book and start on another long-delayed one. In addition to the files and record of his former lab, he also will be curating the records, files, samples, and other materials from the first 14C laboratory of Willard F Libby at the University of Chicago and from Libby’s Isotope Laboratory in the Institute of Geophysics and Planetary Physics, University of California, Los Angeles (UCLA), including the professional papers and files of the late Rainer Berger, as well as samples and other materials from the Mt. Soladad laboratory of Hans Suess at the University of California, San Diego.

131 ERRATUM

There is an error in the heading of Table 1 of McCormac et al., Calibration of the radiocarbon timescale for the Southern Hemisphere: AD 1850–950, Volume 44(3), p 641–651. Both Chilean and Tas- manian wood were included in Table 1, not just Chilean. The correct table title is given below: Table 1 Measurements on Chilean wood samples (Notofagus dombeyi) from AD 1660–1891 and Tasmanian wood samples (Lagarostrobos franklinii) from AD 1896–1981 at the University of Washington. Uncertainties are based on counting statistics. 14C ages are not given for samples later than AD 1958, after which 14C from nuclear testing makes 14C ages meaningless.

133

The University of Arizona Department of Geosciences RADIOCARBON 4717 E. Fort Lowell Rd, Rm. 104 Tucson, AZ 85712-1201 USA An International Journal of Cosmogenic Isotope Research E-mail: [email protected] http://www.radiocarbon.org/ 2004 PRICE LIST

Proceedings of the 17th International Radiocarbon Conference $60.00* (Vol 43, Nrs 2A, 2B and 3, 2001)

Proceedings of the 16th International Radiocarbon Conference 50.00 (Vol 40, Nrs 1 and 2, 1998) INTCAL98 (1998 Calibration issue; Vol 40, Nr 3, 1998) 40.00 Calibration 1993 (Vol 35, Nr 1, 1993) 40.00 Proceedings of the 15th International Radiocarbon Conference (Vol 37, Nr 2, 1995) 50.00 Liquid Scintillation Spectrometry 1994 (ISBN: 0-9638314-3-7; 1996) 20.00 Liquid Scintillation Spectrometry 1992 (ISBN: 0-9638314-0-2; 1993) 10.00 Liquid Scintillation Spectrometry 2001 (ISBN: 0-9638314-4-5; 2002) 60.00 Special offer—LSC 92 , LSC 94, LSC 01 package—save $5.00 85.00 Tree Rings, Environment and Humanity (ISBN 0-9638314-2-9; 1996) 20.00 (Proceedings of the International Tree-Ring Conference, Tucson, Arizona, 1994) SUBSCRIPTION RATES VOLUME 46, Nrs 1–3, 2004 (subscriptions include free online access) Institution (Includes domestic shipping) 190.00 Institution (Includes foreign shipping) 205.00 Individual (Includes domestic shipping) 90.00 Individual ( Includes foreign shipping) 2021236549875605 105.00 Lifetime Subscription—Institutional 2500.00 Lifetime Subscription—Individual 1100.00 BACK ISSUES (except conference proceedings and special issues) Single issue 40.00 VOLUMES 1–9 each volume 40.00 VOLUMES 10–21 each volume 65.00 VOLUMES 22–44 each volume 100.00 Radiocarbon Conference Proceedings 50.00 SPECIAL FULL-SET OFFER—Volumes 1–44 (1959–2002) 1300.00 Big savings. Includes bound copies of 35 out-of-print issues. Take $50.00 off for each additional set.

2004 POSTAGE AND HANDLING CHART Surface mail rates are listed here. Please contact us for airmail or express delivery rates. U.S. Foreign Subscription -- $15.00 Orders must be prepaid. We accept payments by Visa and MasterCard, or by check or money Single back issue $4.00 $9.00 order payable in US$ to Radiocarbon. Bank Book or Proceedings $5.00 $14.00 funds transfers are also accepted. Please contact us for instructions. Full set $200.00 varies Federal tax ID 86-6004791

*Postage will be added; see above chart. Subscription rates and book prices are subject to change. RADIOCARBON The University of Arizona Department of Geosciences An International Journal of 4717 E. Fort Lowell Road Cosmogenic Isotope Research Tucson, AZ 85712-1201 USA

A.J.T. Jull, Editor Tel: +1 520 881-0857 J. Warren Beck, Associate Editor Fax: +1 520 881-0554 George Burr, Associate Editor E-mail: [email protected] Mark E McClure, Managing Editor http://www.radiocarbon.org

August 28, 2003 ORDER OF MATERIALS Radiocarbon Vol. 45, Nr. 1, 2003 PDF files sent on CD-R

Order of File name appearance (in .ps) File type & first author 1. 000_Spine-451 Spine 2. 001_Cover-front Front cover 3. 002_i-ii “Contents” section 4. 003_iii “Editorial Board” section [page iv blank] 5. 004_v “From the Editor” section [will be sent later, p vi blank] 6. 005_vii-viii “From the Former Managing Editor” section 7. 1-7 Article: Yuan [page 8 blank] 8. 9-15 Article: Cordero [page 16 blank] 9. 17-24 Article: Ohkouchi 10. 25-32 Article: Bazarova 11. 33-39 Article: Garcia [page 40 blank] 12. 41-58 Article: Straus 13. 59-73 Article: Michczynski [page 74 blank] 14. 75-80 Article: Guilderson 15. 81-89 Article: Sakamoto [page 90 blank] 16. 91-99 Article: Grottoli [page 100 blank] 17. 101-112 Article: Kristiansen 18. 113-122 Article: Theodorsson 19. 123-130 Article: Bard 20. 131 Radiocarbon Updates section [page 132 blank] 21. 133 Erratum 22. Z-Ad1-IFA Ad: IFA. Start on right side page. 23. Z-Ad2 Ad: Review of Archaeology 24. Z-Ads3 Ads: 1) Geochron [on file at JCI from Vol. 44-1], 2) Catchword, 3) Radiocarbon ad rates, 4) Radiocarbon back issues, 5) Radiocarbon price list 25. Z-Ad4 Ad: Rafter lab (new back cover ad) RADIOCARBON The University of Arizona Department of Geosciences An International Journal of 4717 E. Fort Lowell Road Cosmogenic Isotope Research Tucson, AZ 85712-1201 USA

September 25, 2003 ORDER OF MATERIALS (REVISED FILES) Radiocarbon Vol. 45, Nr. 1, 2003 PDF files emailed

Order of Previous file name appearance (in .pdf) Revised file name & placement 1. 001_Cover-front 001_Cover-front, front cover 2. 002_i-ii “Contents” section, pages i-ii 3. 004_v (not sent) “From the Editor” page v [ page vi blank] 4. 005_vii-viii “From the Managing Editor”, pages vii-viii 5. 25-32 “Bazarova”, pages 25-32 6. 41-58 “Straus”, pages 41-58 7. 59-73 “Michczynski”, pages 59-73 [page 74 blank] 8. 81-89 “Sakamoto”, pages 81-89 [page 90 blank] 9. 91-99 “Grottoli”, pages 91-99 [page 100 blank] 10. 101-112 “Kristiansen”, pages 101-112 11. 113-122 “Theodorsson”, pages 113-122 12. 123-130 “Bard”, pages 123-130 13. Z-Ads3 only two pages of back matter need replacing: 1) “2004 ad rates”, 2) “Advertise in Radiocarbon” RADIOCARBON The University of Arizona Department of Geosciences An International Journal of 4717 E. Fort Lowell Road Cosmogenic Isotope Research Tucson, AZ 85712-1201 USA

September 25, 2003 ORDER OF MATERIALS (REVISED FILES) Radiocarbon Vol. 45, Nr. 1, 2003 PDF files emailed

The Institute of Field Archaeologists was founded in 1982 to advance the practice of archaeology by promoting professional standards and ethics for the conservation, management and study of the archaeological resource. It is the sole body providing professional accreditation for United Kingdom archaeologists.

The IFA Yearbook has been designed to be very useful both for IFA members as a ready source for key specialist products and services, and for the UK’s top specifiers of archaeology work to help them locate just the right specialist for the job. It also features a wide range of colour illustrated technical articles from the top experts.

For Yearbook subscription information please contact Gillian Phillips by e-mail on [email protected]

To receive Yearbook advertising information please contact Gordon Sorensen by e-mail on [email protected] THE REVIEW OF ARCHÆOLOGY

The Review of Archaeology publishes reviews and commentary by distinguished scholars covering a wide range of subjects with the aim of advancing our knowledge of prehistory and stimulating discourse on its various aspects.

C E

H.-G. BANDI (Palaeolithic Art; Arctic) JOHN F. H OFFECKER (Russia, Beringia) DANIEL H. SANDWEISS (S America) OFER BAR-YOSEF (Old World) ARTHUR J. JELINEK (Palaeolithic) DAVID SANGER (Northeast U.S.) PETER BELLWOOD (SE Asia; Pacific) DAVID H. KELLEY (Mesoam.; Epigraphy) RALPH S. SOLECKI (Near East) AMILCARE BIETTI (Mediterranean) MARY ELIZABETH KING (Textiles) ROSE L. SOLECKI (Near East) JOHN R. F. BOWER (Africa) C.C. LAMBERG-KARLOVSKY (Central Asia) MATTHEW SPRIGGS (Pacific; SE Asia) JEFFREY P. B RAIN (Historical; SE U.S.) STEVEN L. KUHN (Hunter-Gatherers) MARY STINER (Zooarchaeology) WARWICK BRAY (Isthmus; S America) YAROSLAV V. K UZMIN (Siberia; Geoarch.) JAMES B. STOLTMAN (Central U.S.) ROY L. CARLSON (W Canada) PAUL LANE (Africa) R. E. TAYLOR (Archaeometry) PAUL A. COLINVAUX (Palynology) SARAH MILLIKEN (W Europe) PAUL TOLSTOY (Mesoamerica) DENA F. D INCAUZE (Northeast U.S.) MICHAEL E. MOSELEY (S America) CHRISTY G. TURNER II (Physical Anthro.) WILLIAM R. FARRAND (Geoarchaeology) HANSJÜRGEN MÜLLER-BECK (W Europe) PAOLA VILLA (Palaeoanthropology) RICHARD I. FORD (Environ. Archaeology) MICHAEL J. O’BRIEN (Mississippi Valley) MICHAEL WALKER (W Mediterranean) GAI PEI (N China) STANLEY J. OLSEN (Zooarchaeology) FREDERICK H. WEST (Beringia) JOSEPH H. GREENBERG (Hist. Linguistics) MEHMET OZDOGAN (Neolithic) J. PETER WHITE (Australia; Pacific) R. DALE GUTHRIE (Palaeontology) D. T. POTTS (Near East) W. R AYMOND WOOD (Great Plains) FEKRI A. HASSAN (N Africa) G. PHILIP RIGHTMIRE (Palaeoanthro.; Africa) WILLIAM B. WORKMAN (NW N America) DEREK A. ROE (W Europe)

Articles in the Fall  issue (Volume , No. ) include two invited reviews in addition to those by Contributing Editors: A. Bietti and A.M. Bietti-Sestieri on Prehistory of Sicily; W.R. Farrand on Geoarchaeology; R.L. Lyman on Zooarchaeology; S. Milliken on Early Hominid Dispersals; S. Schroeder on Mississipian Political Economy; P. Stahl on Revising Amazonian Prehistory; and C.G. Turner II on Human Skeletal Remains in the Southwest.

Subcriptions:  year .;  years .;  years . For a descriptive brochure, subscription information, and a list of available back issues, please write to: Dept. RC Post Office Box  Williamstown, MA  USA

www.reviewofarchaeology.com Reserve for b/w GEOCHRON ad Files in Quark, sent to JCI June 2002 RADIOCARBON is available free online with a print subscription.

Benefits include: • Document to document linking via references • Fully searchable across full text, abstracts, titles, TOC, and figures • Links to and from major Abstract and Indexing resources • Full text searching across multiple journals • Fast download, browsing and printing times • TOC alerting service • Multi-format delivery options including PDF

All you need to do is go to: http://www.catchword.com/titles/00338222.htm

and follow the instructions.

With free software and free support, viewing your journals online has never been so easy.

CatchWord Enquiries: [email protected] ADVERTISE IN RADIOCARBON

REACH YOUR TARGET AUDIENCE EFFECTIVELY & AFFORDABLY

Prices good through March 1, 2003 Display Ads Inside page Inside back cover Back covera $350/issue $500/issue $600/issue $900/volume $1250/volume $1500/volume aReserved through 2003.

Color Ads and Folded Insertsa Color ad (inside page) Folded inserts $1600/issue $300/issue $4500/volume $700/volume aThe prices above are for prepared inserts and ads provided by the customer. Call for estimates on typesetting, design, duplicating and folding services.

Radiocarbon reaches libraries, labs, individuals, and institutions in North America, South America, Europe, the Middle East, Asia, Africa, and Australasia. We can help you reach your target market, wherever it is. We publish three issues per volume year. Advertise in two consecutive issues and get the third at a significant discount (must reserve all three ads at once for discount to apply). We accept full-page and multipage advertisements in black and white or full-color. You may send camera-ready copy, film negatives, or dig- ital files. Or, for a small fee, we will create an ad from your supplied text and graphics.

For more information, please contact Managing Editor Kimberley Elliott at [email protected]. RADIOCARBON $5 Back Issues Clearance Sale

Write in the number of copies Calculate Payment desired for each issue

NR 1 NR 2 NR NR NR 3 NR 4 ____ copies × $5 ea. = $______YEAR VOL. 2A 2B 1980 22 Subtotal: $______1981 23 1982 24 Shipping via surface mail: (Contact us for other shipping methods) 1983 25 Add $2 ea. book for US 1984 26 Add $3 ea. book outside US 1985 27 ___ copies × $____ea. = $______

1986 28 Shipping total: $______1987 29 1988 30 1989 31 Total due: $______1990 32 1991 33 1992 34

Payment & Shipping Address MasterCard Visa Check enclosed Please bill me

Name:

Address:

Fax: E-mail:

For Credit-Card Orders

Card number: Expiration date:

Signature: Phone:

For online Contents see: http://www.radiocarbon.org/Pubs/contents.html

Prices good while supplies last. Mail or fax this form to: RADIOCARBON, 4717 East Fort Lowell Road, Room 104, Tucson, Arizona 85712-1201 USA Phone: +1 520 881-0857; Fax: +1 520 881-0554; [email protected]

Proceedings of the 17th International Radiocarbon Conference $75.00* (Vol 43, Nrs 2A, 2B and 3, 2001)

Proceedings of the 16th International Radiocarbon Conference 50.00 (Vol 40, Nrs 1 and 2, 1998) INTCAL98 (1998 Calibration issue; Vol 40, Nr 3, 1998) 40.00 Calibration 1993 (Vol 35, Nr 1, 1993) 40.00 Proceedings of the 15th International Radiocarbon Conference (Vol 37, Nr 2, 1995) 50.00 Liquid Scintillation Spectrometry 1994 (ISBN: 0-9638314-3-7; 1996) 20.00 Liquid Scintillation Spectrometry 1992 (ISBN: 0-9638314-0-2; 1993) 10.00 Liquid Scintillation Spectrometry 2001 (ISBN: 0-9638314-4-5; 2002) 60.00 Special offer—LSC 92 and LSC 94 package—save $5.00 25.00 Tree Rings, Environment and Humanity (ISBN 0-9638314-2-9; 1996) 20.00 (Proceedings of the International Tree-Ring Conference, Tucson, Arizona, 1994) SUBSCRIPTION RATES VOLUME 46, Nrs 1–3, 2004 (subscriptions include free online access) Institution (Includes domestic shipping) 190.00 Institution (Includes foreign shipping) 205.00 Individual (Includes domestic shipping) 90.00 Individual ( Includes foreign shipping) 2021236549875605 105.00 Lifetime Subscription—Institutional 2500.00 Lifetime Subscription—Individual 1000.00 BACK ISSUES (except conference proceedings and special issues) Single issue 40.00 VOLUMES 1–9 each volume 40.00 VOLUMES 10–21 each volume 65.00 VOLUMES 22–42 each volume 100.00 Radiocarbon Conference Proceedings 50.00 SPECIAL FULL-SET OFFER—Volumes 1–43 (1959–2001) 1200.00 Big savings. Includes bound copies of 35 out-of-print issues. Take $50.00 off for each additional set.

*Postage will be added; see above chart. Subscription rates and book prices are subject to change. The test of time Accelerator Mass Spectrometry Radiocarbon Dating

Rafter provides: At the Rafter Radiocarbon Laboratory we have been F Prompt reporting times successfully meeting the test of G never more than 8 weeks time for more than 50 years. F Competitive prices G NZ$ 820 per sample The Rafter Radiocarbon F Laboratory has an international Stable isotope corrections reputation for accurately G d13C measurement at no extra charge dating a wide range of organic F Additional services materials, sediments, textiles, G 3 week express turnarounds bone, ivory, paper, wood, G enhanced precision measurements parchment, charcoal, shell, G 14 radiometric C dating for large samples foraminifera and peat. G bone C:N analysis:- %C, %N, d13C,d15N G direct dating of pollen At Rafter we understand what G X-ray diffraction for shells our clients expect: accurate G calibration assistance dating, at competitive rates G consultancy services and superior turn around times and service.

For more information, visit our website www.RafterRadiocarbon.co.nz

or contact us at: Rafter Radiocarbon Laboratory Institute of Geological & Nuclear Sciences 30 Gracefield Road, Lower Hutt, New Zealand %: +64-4-570 4647 Fax: +64-4-570 4657 Email: [email protected] ADVERTISE IN RADIOCARBON

REACH YOUR TARGET AUDIENCE EFFECTIVELY & AFFORDABLY

Prices good through March 1, 2004 Display Ads Inside page Inside back cover Back covera $350/issue $500/issue $600/issue $900/volume $1250/volume $1500/volume aReserved through 2003.

For more information, please contact Managing Editor Mark McClure at [email protected].