Abstract IwmoI O' ? ArchaeologIcal SCIENCE Journal of .... rchllWlogica! Scienc:e 34 (2001) 28-31 1l1.1):lIwww.elsevier.co.nllocatelja~ Mixed results of seven methods for organic residue analysis applied to one vessel with the residue of a known foodstuff H. Barnard a.*, S.H. Ambrose b, D.E. Beehr b, M.D. Forster C , R.E. Lanehart d, M.E. Malainey ' . R.E. Parr ' . M. Rider ' . C. Solazzo h. R.M. Yohe II ' ? Cmu" /"lfiMt oj Ior('hot%l/Y. Univusity oj Coli/ornia. Los Io nlltln. P.O.?Bo.r 9515/0. Los 1o"lIells. CI. 90095, USA ~ Deportme,,' o/II",hropolUII),. U"irtrlity '" mi"o# 01 Uroo/W, MJ7 SOII/h Murh,...?s llvellile. Urha flO. IL 61801. USA < Dtpartmelll 0/ "'rhatu/ol/ica/ Sdtllcts. Unil'trsi f)' 0/ Brad/ord. Brad/ord. B07 lOP. UK d AIIII"u(IOloIIY Otp""m,,,,. U"il?trlily 0/ SOlllh Florida, 4202 ?all Fowler ",'efllle. Tompo . FL lJ/l20. USA < Deporlmtlll 0/ "IIIllrropology. 8rOlldo" U"i"ersity, 270 181h SlrUI. Manirooo. R711 6,019. Conodo f Loboro/Qry 0/ Archor%llic-Il l Sdtnres. Cali/ornio S/ole Unil'tfsiry. Boktnfitld. 9001 Sioddolt IliXh .. ",y. BoJ:trs/rtlrl. CII 93311, USA ? Deportmelll o/IIIIIhrop%IIY. U"i .. trsity oj "'i: ollo. P.O.?Bru 210030. Twsoll. liZ 8j72I. US,," ~ Smifhsonion Centn /or MOfuiall Rtltarrh and Edllcofioll (now lilt MU ftum Co"Strvo/;on Inslillllt). 4210 SiI,'" Hill Rood. Suil/alli/, MD 1074/1, USA Received 11 January 2006; received in rev ised form 18 March 2006; accepted 20 March 2006 Several methods of archaeological organic residue analysis were applied to a single unglazed and unseasoned ceramic vessel that had absorbed rcsidues of heated camel milk. Sections of the wall of this vessel were sent to eleven archaeological laboratories. Seven reported their results before the identity of the residue was revealed, during the 70th Annual Meeting of the Society for Amcriean Archaeology, Methods in? c1uded stable carbon and nitrogen isotope ratio analysis. protein analysis and lipid analysis. These laboratOf)' t~hniques provide a biochemical analysis of the residue in a ceramic matrix, the archaeological interpretation of which can be rather difficult. The exact source of the rcsidue was not identified by any laboratory. but it is evident that residue analysis can provide valuable information, especially when combined with addi? tional archaeological aod historical da ta. We therefore support a close cooperation of those worki llg inlhis field to develop ilto its fu lilloteotiat. ro 2006 El sevier ltd. Al l rights reserved. Ke) ... ?ords; Carbon isotopeE; Calml milk; Come/us dr(Hfltdariul; Cer. mic anal)'~i.; Diet reCOnSlrltClion; Electrophoresis: Fally acids; GaE chromJIOGraphy: Lipi d~; Mus spec!rometry; Nitrogen i ~olopes; Prole in~; Residue analYlii! I. Introductio n On 3 1 March 2005 wc organized a ~ympos ium entitled 'Theory and Pract ice of Archaeological Residue Analysis' f56J. With assistance of Dr. Jctmer Eerkcns (University of Cal ? ifornia, Davis) and Dr. Ran Boytner (University of California, Los Angeles), this symposium took place in Salt Lake City. duri ng the 70th Annual Meeti ng o f the Society fo r American ? Corresponding aUI""r. Tel.: + t 3 10 261 5550; fu: + 1 310206 4123. t?? IIIoil addr flS; wendrich @l bllfn;r.rd .ul (H. Barnard). UHL: hlll';/l ... ww.han,ard.AJn'al1~ 1 .hI11l1 0305?44031$ ? ~ce frolll mailer CI 2006 Elsevier tid. All rigllt~ reserved. doi: I o. t 0I61j.jas.2006.01.0 to Archaeology. as sessions 21 and 46. The proceedi ngs of these will be publi shed in the British Archaeological Repons, Inter? nat ional Series. The participants were asked to prepare presen? ta tions on their own research. and to report on the ir analysis of a res idue of a foodstuff recently eooked in a new ceram ic ves? scI. Eleven agreed to partake in the latter and were sent a seg? ment of the wall of a Yessel in which camel milk had been cooked. The accompanying letter and model report sheet did not provide infonnation on what was prepared in the vessel, nor did it present a list of possibil ities. Thi s made the analysis more challenging than is typical of archaeological sett ings as the provenance, shape and d ll te of a pot usua lly offe r important elues to its fo rmer use and contents. Seven partic ipants (a lmost II. Bamart/ el a/. I Jownal of Archaeological Sriem'e J4 (2007) 28- J7 29 2/3 ) in this blind ' Round robin' fil ed a report before the meel" ing where the source of the residue was revealed. These re? ports will be discussed here, preceded by some data on camel milk and followed by a discussion on the significance of our findings for the practice of archaeological residue analysis. 2. Creation of the residue In December 2003 about 200 ml of fresh camel mi lk, ob? tained at the camel market in Daraw (Egypt), was placed in a new unglazed earthenware bowl, purchased in Luxor. This was topped off with Baraka mineral water (total dissolved solids 430mg/I), wrapped in aluminum foil and allowed to sit at room temperature for 24 h. The next day the asscmblagc was put in a gas oven, heated to approximately 200 ?C for one hour, left to cool for four hours, again cooked for an hour and left at room temperature for another 24 h. The vessel was then emptied, rinsed with cold water and air-dried for 10 days. Fi? nally, the vessel was rinsed with cold water to remove the fun ? gus, air-dricd for 24 h, stored in a scaled plastic bag and transported to Los Angeles, California, In May 2004 Ihe vessel was machine-cut in 12 pieces, one being the base, after which the I I wall fragmcnts were distributed among the part icipants of our blind 'Round robin' . ), Camels and camel milk Millions in Asia and Africa rely on camels (the one? humped Came/us dromedarius and the two-humped Came/us bactrianus) for transportation, milk, meat and leather, espe? cially in areas that arc too hot and dry for other large mammals such as cattle and horses ]6,18.32.47.49]. A healthy lactating camel produces 5-15 kg milk daily for a period of 9- 15 months [4, 17.35J. The milk has a relatively high pH (6.5-6.7) and high concentrations of vi tamin C and niacin ] 16,20.461. The absolute and relative composition of the milk depends on the fodder and the stage of lactation 155], like in other mammals, but also on the state of hydration of the animal. In camels the latter can vary greatly as part of their adaptation to life in an arid environment [53J. Among other things, lack of drinking water will cause an increase in the mineral contcnt of the milk and a dccrease of fat, lactose and protein [27,28,5 1,52] . The fatty acid composition of the fats in camel milk is also rather variable (Fig. I). In fresh milk the fats arc suspcnded in micelles (globules), about 300 f.lm in diamcter, in which they appear to be bound to pro? teins [21,43,50J. The fatt y acids found in the ceramic matrix of a vessel that was used to contain or process camel milk must have been those present in the milk or the products of oxidation, or other reactions, of these compounds. Their absolute and relative abundance will depend on many factors, including the affinity of the ceramic matrix for each of the various molecules, the stability of those molecules over time and the efficiency of the extraction and dctcction techniques em? ployed, Givcn what is known about the fatty acid signatures of fresh camel mi lk and organic residues lypically recovered in archaeological contexts, the saturated fatty acids C16:0 (palmitic or hexadecanoic acid), ClS:0 (stearic or octadeca? noic acid ) and C 14:0 (myristic or tetradecanoic acid) were anticipated in the comparatively fre sh residue of our blind ' Round robin'. Fair amounts of the mono-unsaturated CIS:I (oleic or octadecenoic acid) and CI6:J (palmitoleic or hcxadccenoic acid), and their oxidation products (like di? carboxyl ic acids), wcre also expected (Fig. I). These fatty acids arc com mon in fats and oils of vegetable and animal origin and their presence alone is unlikcly to allow F.tty ,eld eomposltion Ig/100 g) of e.mtt milk 1.1 C1 f .. . h .1 .1. 1.89 Fig. t. Comparison of seven fatly acid signawrc~ of fresh camel milk I I ,2,21 .22.26.33A IAIi.50] (aflcr I,UIl. 30 H. B(JrnPTd tl ul.1 hmrnul of lI~hato/D/lic'a/ SritnC'f' J4 (1007} 2R-J7 identification of an unknown residue such as thai of camel milk. Making such an identification docs require addit ional infonna lion and further manipulation of the data . Another analytical technique, which could be used inde? pendently or in combination with the identification of specific compounds, e ntai ls the determination of the rat ios of the stable isotopes of carbon (' l C and DC) and ni trogen (14N and I ~N ). Families of plants have their own specific preference for onc isotope over another and isotope ra tios in living matter are therefore dissimilar to those in the environment. In the New World, for ins tance, maize (Zea mays) employs the C4 path? way for phOiosynlhesis which causes higher 1lC/12C ratios. while introduced European cereals employ the C] pathway, which produces lower IlC/12C ratios [44J. This difference in BC/llC ratios is carried up the food chain: herbivores feedi ng on C4 plants. and carnivores feeding on such herbivores 15,421. have higher Dcl2c rat ios than those feeding on C] plants. Egyptian camels typi cally feed on both C) (most grasses and trees) and C4 plants (sorghu m, sugar cane). Their milk contains 3-5 g protei n (mostly casein) per 100 g 11.2.17.35. 43.461. which should leave small amounts of nit rogen in its residue. The analysis of protei ns such as casein would potentially enable a secure identification of the source of mRny archaeo? logical residues (Fig. 2). as many proteins are species?specific [24,25). Ancient proteins have been isolated from archaeolog? ical samples with a variety of techniques 13.8 .9. 19.36,37, 48.541. That this avenue has not yet been full y explored can Ph. nyl~l~n l n. : 4.01 Slrlne: 5.39 LY5.n.: 6.53 be attributed to the facts that proteins are relat ively difficul t to handle in the laboratory and arc expected 10 have denatured and fractured over time. especially after being heated. Further? more, not all naturally occurring proteins have yet been fu lly analyzed, which would be necessary for com parison with archaeological residues [24,251 . Spec ific antibodies are readi ly available for only a limited number of proteins, and may not react with severely dCnlllurcd or damaged proteins, although promising results have been obtained in particular cases 18,91. Counter or cross-over electrophoresis (C IEP), also employing the principles of antigen-antibody reactions. has been applied on archaeological materials [3,31,36,37.48,54 ), although some hnve expressed scepticism as to the reliability of thi s approach 17,\3.15,23). The rapidly evolving field of prot comics may soon develop new techniques that can also be uscd for archaeological residue analysis, especially with an increased cooperation between biochemists and archaeologists. 4, Results or our blind ' R()und robin '; s table is()topes Laboratory J removed the outside surfaces o f the sherd, to avoid possible contamination, after which a tolal of 12 sam? ples were obtained at evenly spaced inlervals from the rim to the base or the vessel segmenl, six from the interior and s ix from the exterior. Aliquols of Ihe ceramic powder, weigh? ing an average of 4.9 mg, wcre manually compressed in ti n foi l capsules and combusted in a Carlo Erba NA2500 r Oil",",,,: 0 .1 I , Glllt~mlc aCid: 21 .26 ProUn.: 11.62 Leucln.: 10.89 A5panlc ~ e 'd; 7.28 Amino acid composition (g/100 g) of camel milk casein (Farah and Ruegg,1989) I' i&. 2. The amioo acid 'ign~lulc of elscin in frc~h calIlCl mill: (after [21 n. II. BarMrd tl 01. f }ourflDI of IIrrNltolDf(ir of the three samples closest to Ihe vessel base range from 9 to 27. These arc higher than the C/N rat ios of pure proteins. but arc substantially lower than those of pure carbohydrates and li pids, thus indicating a mix of protei ns and carbohydrates. As only one foodstuff was cooked in lhis vessel it was likcty cither a protein-riCh seed, or a whole milk product. The moderately high OIJC and al5N values excl ude 13C-depleted C.1 and uC-enriched C4 plant foods. )2 If. Burnard tI Ill. I }ourm;jl of Arrhat%giral Sritnrt 34 {lOO?, 28-37 o ~+--------.---------r-------' o 0.' 0.' 0.3 wt "N rig. 5. Change in nitrogen conce ntrations of lhe abUlr~d residue in powdered ~amples from six po$itions below the interior of the vessel rim. The strong co-variance of the interior and exterior ~llC values of the five samples closest to the rim suggests that the upper residues were readily absorbed in the ceramic rna? trix. The upper matrix may, however. nOI have had the same capacity for absorption of higher density proteins lind carbo? hydrates thai predominated near the basco The systematic patterns of carbon and nitrogen concentrations and isotopic composition of the interior and exterior of this vessel were entirely unanticipated. However, these pallerns are consistent with the preparation of a single foodstuff Ihat had contained lipids and a C/N ratio higher than that of plants but lower than that of pure protein. Variation in carbon and nilrogen isotopic and elemental composition within a vessel used to cook a single substance can be substantial due to systematic differences in the isotopic composition and densities of lipids, proteins and carbohydrates. Control over the sample position wilhin the vessel is therefore essential for the ade? quate interpretation of carbon and nitrogen elemental Dnd isotopic composition of absorbed organic mailer in potsherds. The simple sampling strategy described here, combined with analysis of carbon and nitrogen elemental and isotopic com? position, can provide a considerable amount of useful infor? mation about the foods prepared in porous ceramic vessels. o 0 '------ 0 0 / / , 0 o " " " " 10 80 Alomlc c/N Fig. 6, Change in atomic C/N ratios of the abJorbed residue in powdered ~a.nplt. from six positions betow the interior of tilt vesstl rim. 5. Results of our blind 'Round robin': proteins Laboratory A placed the sherd in a plastic dish with 0.5 ml ammonium hydroxide (5%) and sonicated for 5 min. Both dish and contents were then placed in a rotating mixer for 30 min. The resulting solution was transferred into a sterile plastic vial and stored at -20?C. About 3 ,t! of this ammonia solution was later transferred into a well punched in an agarose gel in a Helena Laboratories Titan Gel electrophoresis chamber and paired with a second wcll with 3 J.ll antiserum. A control positive was prepared in another pair of wells after which an AC current of 105 V was passed Ihrough the gel for 45 min. This caused sample and antiserum to migrate, antigens towards the anode and antibodies (antiserum) towards the cathode, and come into conlact. If there is protein in the sam? ple that corresponds with the antiserum, an antigen-antibody reaction will occur resulting in the protein precipitating out in a specific pattern. After the run was completed the gel was pressed and dried. The dry gel was immersed in a Coo? massie Blue R250 stain for 3 min and then destained in II solu ? tion of ethanol, distilled water find acctie acid (5:5: I, v/v) until the background was clear lind any positive responses visible. Sterile equipment was used throughout the analysis. The IIntisera used in this study included agave, amaranth, bear, bovine, cactus, cat, cedar, chicken, CapparidaCt!a~, Ch~? nopodiaceac. Compositae, deer, dog, guinea pig, Graminae, legume, pine, rabbit, rat and sheep. All the animal antisera are from Cappel, purchased from MP Biomedicals, while the plant antisera were produced by the Department of Biological Sciences at the University of Calgary. All antisera arc poly? clonal: they recognize epitopes of closely related species . For example, anti-deer serum will react positively to other members of the family CCrI'idae such as elk, moose and cari? bou. Commercial antisera manufacturers provide product sheets listing cross reactions; the antisera from the University of Calgary arc tested against a full suite of olher species to ensure that there arc no cross reactions. This method of analysis is known as cross-over (or counter) electrophoresis (CIEP). Prior to the introducti on of DNA fingerprinting this technique was commonly used in forensic laboratories to identify residues from crime scenes. Minor adaptations to the ori ginal method were made following proce ? dures used by the Centre for Forensic Sciences, Toronto, and the Royal Canad ian Mounted Pol ice Serology Laboratory. Ottawa (451. The solution used to remove residues, ammo? nium hydroxide, has proven to be the most effective extractalll for old lind denatured bloodstains without interfering wilh subsequent testing 11 2,34]. No trace proteins were detected in the sherd sent for analysis. This may be the result of the degradation of animal proteins that result from cooki ng. Experiments subsequent to this study have demonstrated that an imal proteins, of known origin, heated to 100 ?C for 20 min yielded no positive read? ings using CIEP. This observation has important implications for the applicntion of this particular form of residue analysis on cooking vessels or objects from hearth. Another issue to consider is that many of the taxonomic groups for which H. Barnard ~I 0/.1 Jill/mol II! Ar('hOM/ogico/ Scitnr~ 34 (2007) 28-37 II antisera exist, like those used in this study, would not be ex? pected to react to the residues in the 'Round robin' (with the possible exceptions of bovine and sheep). 6. Results of our blind ' Round robin': lipids Laboratory B ground a small piece of the sherd in a mortar and pestle. The resulting powder was transferred into a sterile test lUbe with 8 mt acetonitrile (ACN). The mixture was son? icated and centrifuged. after which 5 ml was pipetted into a second test tube. The solvent was evaporated from this test tube by gentle heating under a stream of nitrogen. The dry res? idue was taken up in 100 ~d ACN and treated with 30 ~Il trime? thylsilyl-N-methyl-trill uoroaeetamide (MSTFA) . About I J.l1 of the derivitized sample was sandwiched in air and injected. pre ? ceded by 1 J.l1 of ACN. into a Varian GCIMS instrument. Two different pieces of the same sherd were analyzed. by comparing CI8:OJCI6:0 and CI8: [JCI6:0 ra tios. One sample showed a C I8:0JC I6:0 ratio of 0.72 and a CI8:lIC I6:0 ratio of LIS (Fi g. 7). The second sample showed ratios of 1.03 and 0.32. respectively. Fatly acid amounts in known foodstuffs were obtained from the USDA Nutrient Data Laboratory web? sitc and used as a comparative data set. The peaks of all the fatty acids that were tested for (CI2:0, C I4:0, CI6:0, C I6: 1, C I8:0, CI8: 1, C 18:2 and C20:0) were compared with those in the known foodstuffs. The only peak that differed dramat? ically was that for C 18: 1 (oleic acid). This may be due to the oxidation of this unsatu rated fatty acid, which may have also caused the difference in ratios between the two pieces tested. The chromatogram containing the higher amount of C 18: I was compared with the data of known foodstuffs. The results were tentatively interpreted as indicative of the residue of some animal product and most likely veal, eggs or goat milk. Laboratory C received the snerd broken into two vertical strips. The upper portion of one strip was selected as the ,. " , .. ~ 1.2 ~ 0::: 1.0 ? ~ 0.8 ~ 0.6 0.' 0.2 D.D --I ' I- I-: 11 ,0neo 01' :1116:0 f-- f-- ~ f--l- f--L.- ."0 .;>~ ~~ ? ~. .,~ lfl b'::' "Q,t; c,'11 ~ ... .;:."~ ~~ ,0 - f-- - f-- - f-- - -f--- f-- f-- f- B f--b Fig. 7. Graphic repre:;cnlalion of lhC C I8:0iCI6:0 ,nd C 18: tIC t6:0 ralios in two samptes of lhe 'Round r<:>bin' . cSlablished ill Laboratory B. cotnl"'red 10 those in six foodstufb . 't ime 0' sherd. The remainders of tne pieces were stored in an oven at 75 ?C. After six days the lower portion of the first strip was removed from tne oven and stored in a freeze r at _20?C. Tne same was done, after 12 days, with the upper portion of the second strip. Residues were then extracted from the 'time 0' sherd and from those stored in the oven for six and 12 days. Contaminants were removed by g rinding off the surfaces after which the sherds were crushed. The re? sulting powder was mixed with 30 1111 of a chloroform and methanol mi xture (2:1, v/v) and sonicated (2 x 10 min). Solids were removed by filtering into a separatory funne l. The solvent mixture was wasned with 16 ml ultra-pure water and left uatil it separated into two phases. The lower chloro? form-l ipid phllse was transferred into a nask from which the chloroform was removed by rotary evaporation. Any remain? ing water was removed by evaporation with 1.5 ml benzene. The dry residue was transferred into a vial with 1.5 ml of the chloroform and methanol mixture and stored. A 200 J.ll sample of the solution was dried under nitrogen and treatcd with 6 ml 0.5 N anhydrous hydrochloric acid in methanol. Af? ter cooling 4 ml ultra-pure water was added. The fatty acid methy l esters (FAMEs) were recovered with 3 ml petroleum ether and transferred into a vial. The solvent was removed by heat under a gentle stream of nitrogen. The dry residue was transferred into a GC vial with I ml iso-octane. Analysis was performed on a Varian 3800 gas chromato? graph fitted with a flame ionization detector. Chromatogram peaks were integrated using Varian Star Chromatography Workstation software and identified through comparisons with several external quali tative standards (NuCheek Prep, Elysian. MN). To identi fy the residue the relative percentage composition was determined, firs t with respect to all fatty acids present in the sample, and second with respect to the ten fatty acids utilized in the identification cri teria: CI2:0; CI4:0; C I5:0; CI6 :0; C I6:1; C I7 :0; CI8:0; C I8: 1w9; C18:lwll and C I8:2. Medium chain fallY acids represcnt the sum of C I2:0, C14:0 and C I5:0; while C18:1 is the sum of all isomers. It must be understood that Ihe identifications given do not necessarily mean that those foods were actually prepared because different foods of sim ilar fatty acid compo? sition and lipid content can produce similar residues. Significantly more fatty acids were recovered from the up? per portions than of the lower porlions of the sherd and, con? sequently, these provide the best information about the residue. The characterization is based on the relative fatty acid composition of the residue extracted from the upper por? tion of the second strip. Due to the low level of unsaturated fallY acids in the residues, it is possible to establish decompo? si tion trends after on ly 12 days of ovcn storage. The sum of medium chain fatty acids exceed.~ 30%, while the C 18:0 and C 18:1 isomer levels arc low. By applying published criteria it is possible to say that the partially decomposed residue is typical of decomposed plants 139.401. This category includes low fat-content plants such as roolS or tubers, greens as well as certain berries and seeds. North American foods known to produce residues with high levels of C 14:0 when boiled arc biscuit rOOI and Chenopodium seeds. 1/ , Barnard el ai, I }aurnal of Itrrhaea/allira/ Srietwe 34 (2007) 28-37 Laboratory G ground off about 0. 1 g of the internal surface of Ihesherd, mixed the result ing powdcrwi th I ml of a chloroform and methanol mixture (2:1, vlv), sonicated and centrifuged. The supcrnatant was decanted and dried under a stream of nitrogen. The dry residue was treatcd with 30 III N,O-bis(trime? thylsilyl)trifluoro-acetllmide (BSTrA) and about I III of the derivatized sample was injected into a Hewlett Packard .5972 CG/MS. Nex t to a trace of C I2:0, the sample appeared to contain a number of free fatty ac ids: C14:0, C J6:0, C18:0 and C 18:1 as well as the methyl-esters ofC16:0 and C18:0 (Fig. 8). Nei? ther short chain fatty acids nor poly- unsaturated fatty acids were identified. No attem pt was made to dctennine lhe ratios of the frcc fa uy acids but C16:0 appeared predominant. There were traces of odd carbon number fatty acids, but these were in low abundance. Cholesterol was present in abundance along with cholesterol oxidation products and a trace of squalene. No phytosterols or alkanes were found. No terpcnoids and no wax esters were identified, indicating that the sou rce most likely did not contain a resin or a W8Jt, On the basis of these results it can be suggested that the residue is of animal origin, the absence of short chain fally aeids pointing to a non-dairy source like veal or eggs. This identification should be confi rmed by saponificat ion of the sample, hydrolysis of the ae)' lglycerols result ing in glycerol and fallY acid salts, fol? lowed by an analysis of the resul t by gas chromatography combustion isotope ra tio mass spectrometry (GC/C/IRMS) as d iscussed by Laboratory J above 144]. Laboratory H ground the sherd in a mortar and pestle. Lipids were ex tracted in a chloroform and methanol mixture by sonification. This was centrifuged and decanted after which the solvenlll were evaporated under nhrogen. The dry residue was treated with BS .. ,.A, sapon ificd with sodium methoxide in methanol and methylated with a BFr methanol complex. The sample was then analyzed in a GC- J7A Shimadzu Chrom-Perfect gas chromatograph, with a 30 m Agilent DB ? .5 column raised 5 "C/min from 40 "c to 280 "c, after a split injection, No activity was found in thc reeeivcd sherd. Laboratory K cut the sherd into four pieces, two of which were crushed into a fi ne powder using a mortar and pestle. "" Fl~, 8. Chromatogram showing the lipid profile of Ihe 'Round robill' ~s e$(pb? lishcd by Labor3(ory G. ME: methyl-esler: MAG: monoacylglyceroJs; DAG: di ac ylgtyccrots: TAG: triaeytgtycerols (cr. (1i~. 9 ). Lipids were extracted by sonification from .500 mg of the powder in 2 ml chlorofonn/methanol (2: I, v/v) after which the solvents were evaporated undcr a stream of nitrogen. .50 III tetratriacontane (CHy(CH2)n-CH) was added as inter? nal standard and the dry residue was then treated with BSTFA with 1% trimethyl-chlorosilane (TMCS). The lipids in another .500 mg were saponified with 0.5 ml 2 N KOH, acidified with HCI and extracted with diethyl-ether. The solvent was then evaporated under a stream of nitrogen and treated with BSTFA with 1% TMCS after the internal standard was added. The derivatized residues were analyzed in a Hewlell Packard 5890 series II gas Chromatograph (HTGC) and a Hewlett Packard .5890 gas chromatograph with a 5972 mass sclecti ve detector (GClMS). The HTGC was used with a OB - IHT fused-si lica capillary column (1.5 m x 0.32 mm, 0. 10 ~m). The HTGC oven temperature was initially set al.50 .. c and in? creased with 10 "C/min to 380 "C, at which temperature it was held for another 5 min. The GClMS was equipped with a fused-silica capi llary column (30 m x 0.25 mm, 0.2.5 11m). The GC oven temperature was programmcd to start al .50 ?C, to increase with 10 "C/min to 340 "C and fi nally to be hcld at 340 ?C for 10 mi n. The analysis of the total lipid extract. by HTGC, s howed the presence of tri acylglycerols (TAGs) with a total carbon number between 44 and 56, the main consti tuent having 52. Afte r saponification and GC/M S. the main fatty acids ap? peared to be C I2:0, C I4:0, C16: I, C I6:0, CI S: ! and C I8:0. The C I6:0/C I8:0 ratio (PIS ratio) for one pari of the sherd was 3.17.5 by HTGC, and 5.741 by GClMS. For the other part of the sherd the C I6:0/C I8:0 ra tio was 3.09 by HTGC. The distribution of triacylglycerols indicates a residue of ani? mal origin. A CI6:0/CI8:0 ra tio of around 3 rna)' be ex plained by the presence of egg. It was therefore concluded that tin an? imal product was the source of the residue. Analysis of samples taken cight months later, using another HTGC (an Agilent Technologies 6890N) with the same col? umn, showed some interesting di fferences even though the sherd had been stored in a refrigerator (Fig. 9). All fatty acids except C 18:0 had decreased, while the most common triacyl? glycerol now had a total carbon number of 48 instead of .52. It seems that trincy lgJycerols conta ining mainly C18:0 had hy? drolyzed to form free C18:0 and monoacylglycerol C I8:0. The new CI6:0/C I8:0 ratio obtained after saponification appeared very low compared to the fi rst ratio, which is concor? dant with an increase in CI8 :0. Other rat ios, like C I6:I/C I8:1 and CI6:0/CI8: I. showed little variation after eight months. It therefore seems li kely that C18:0 was released, probably because of the temperaturc chunges to which the sherd was exposed. 7, Discussion The fi rst conclusion of our blind ' Round robin ? must obvi? ously be thll t none of us could pinpoint the source of the residue. This may seem disappointing but should be hardly surprising tiS our methods sti ll onl), allow us to look at a small selection of molecules and not directly at foodstuffs II . BUTltOrd r l 01.1 Journal of Arrlwrolulical Sritnrt 34 (2007) 28-37 " "" "0 "" "" "" .. 0 ? I'ig. 9. Chromatogram showing the lipid profile of tile 'Round robin' after eight months of rdrigcr~lcd i tolage as eSlablished by Laboratory K. Tile inter? nal standard is IClflIlriaconlanc; MAG: monoaeyJgJyccrols; DAG: diacylgly? (trl)ls; TAG: Inacylglycerols (ct. Fi~ . S). themselves. In the reduction from camel milk to fauy acids too much information is 10SI and it is impossible to retrieve all o f this. The driving motive behind Ihe 'Round robin' was to get an overview of the methods currently available \0 rccover some of this information in order LO arrive al archaeo\ogically relevant conclusions . We purposefull y include the negative results of two of the laboratories as the trend to only publish successful projects. though fully understandable. has contrib? uted to a biased representation of the state of the ficld and unrealis tic expectations, The second conclusion is that, with the current technology, it will only be possible in certain cases to assign archaeological residues a source more specific than 'mi lk of an hcrbivore' or ' low fat-content plant' , And it will rarely be feasible \0 infer such determinations direc tl y from the chemical analysis alone. Usually additional archaeological lind historical in fonnat ion combined with data from the analysis of thc residucs of, pref? erably art ificially aged, known foods tuffs is required. A residue of camel milk, for instance, need nOt be expected in vessels from North America or in pOlS from Pharaon ic Egypt. That res? idue analysis can contri bute to this assemblage of infonnation is clear from the over view of our interpretations in Table I. This illustrates the neccssity of combining the chemical data with archaeOlogical and historical information. Laboratory C, for instance, specia li zed in the study of North American Table t Summary of the resutt. of ou r anaty~i5 of the residue in the ' Round robin ' "', Techn ique Result A CtEP (20 ~nti M: IJ ) No proleill5 delcclw hunter-gatherers who did not partake in dairying activities. In order to characterize their archaeological residues, usi ng fallY acids, a collection was made of fo odstuffs from that region and residues were prepared and subjected to oven storage 10 simu late the effects of long-term decomposition . As Labora, tory C was under renovation from spring 2004 to spring 2005 the preparation of new reference materials, relevant to the pos? sible residue in the circulated sherd. was not possible. In our ' Round robin' comparisons between unknown resi? dues and those of known foodstuffs were made in two manners that are different but not mutually exclusive, Some residues preserve 'biomarkers', which are more or less specific for cer? tain classes of foodstuffs . Such molecules can be alkaloids (such as caffeine), steroids (such as cholesterol ), lerpenoids (a large diverse group of mostly polycycl ic compounds synthe? sized by plants 110,30]), but also fatty acids like C22:1 (erucic or docosenoic acid) and phytanic (3,7, 1I,1 5-tetramethyl-hexa? deeanoie) acid 1291. Respectively, these compounds naturally occur in coffee, animal products, certain families of plants (the terpenoids arc well?studied and can be very specific 110,301), members of the Bl'ossicoceoe (Cruciferae) family (such as mustard, nasturtium or rape seed) or fish and finally, in the produCIS of ruminanlS or fish 1291. As is evident from the last two examples the same molecule is sometimes consid? ered a biomarker for two apparently unrelated classes. Some compounds can mark modern contamination of the sam ple. This is obvious in the ease o f man-made organic molecules like phthalates, in plastics to keep them flexible. but must also be considered for naturally occurring compounds like anthraquinones, commonly used as dye in textiles and paper, or 13-docoscnamide (erucamide), coaled on many plastic objects to prevent them from sticking together. Proteins arc inherently more specific than the relatively simple molecules mentioned above, and would theoret ically bebcller biomarkers, but their identification from an archaeological context is slill problematic in many cases. As some residues will not contain or preserve detectable concentrations of any biomarker, protein or otherwise. this approach docs not always yield helpful results. The second way to match unknown with known residues is by comparing the I1Iti05 of the abundance of common fatty acids. Results can be someti mes be obtained with a simple CI6:0/CJ8:0 rat io (PIS ralio) and sometimes after the IWO? dimensional plotting on a double logarithmic scale of C 16: II C18 :1 versus (CI 5:0+C17:0)/CI8:0 1141. Such ral ios tcnd tnte.-pretation ProteillS dclllldcd .rtcr cookillg1 B GC/MS ClB:0fC16:0 '" 0.72- 1.03. CIB:11C16:0 : 0.32- 1.\ 5 Veal. egg 01 goat mit !.: C GC Medium ellaill fatty acid~ > 3{)/I,.low CtB:O and Ctl:1 ho mer tcvct$ Dceompo~d root5. tubcu or berries G GClMS CI4:0. CI6:0, C t8:Q, 18:1 Illd cholesterol Veal Or cgg H GC No activity J [RMS b'le", 19.2 ( c ~t.) - 22.3%, (inc.) bl ?'N ? 5.08-6.00%. Mit!.: of an herbivore K IITGC and GC/MS lriacylglyeerols (44-56 C? atolll ~), e 16:OIC t8:0 = 3.175-5.741 Animal product J6 //. Barnord tl 01. I Journal of Mehalological Scknft 34 (2007) 28-37 10 bring the residues from the same class of foodstuffs together rather well, bul often fail to fully separate the different classes. The fact thai these ratios may furthermore change over lime. as different fauy acids oxidize at different rales, further com? plicates the application of this method. One way to approach this is be monitoring suc h changes in known residues in sherds that arc stored in an oven to simulate the effects of long-term decomposition [38 ]. The previously mentioned lipid bio ? markers and fatly acids are usually found with the same ana? lytical techniques, which enables the use of both methods on the same datase!. Th is approach wou ld greatly benefit from a database, preferably accessible through the Internet, in which the raw data on residues from known sources is stored togelher with detailed information on the methods used to obtain these data, The determination of food types by stable carbon and nitro? gen isotope ratios is hampered by the similarities in isotopic composition of food Iypes within a few broad classes such as C) or C4 plants, or animals feeding on these classes of plants. As with other methods, additional archaeological and hi storical data on the studied material benefits the unalysis. This technique rcquircs specific instruments and combining it with those discussed above demands additional samples, time and funding. One way to cope wilh this would be coop? eration and the sharing of data, combined with the analysis of more known samples to evaluate cross-laboratory variation. Perhaps our most important conclusion is that such is nOl only necessary but also possible. Othcr issues in need of attention by those working in the field include Ihe fonnation processes related to organic resi? dues. It is always assumed that the encountered residues arc related to the food that was once present in the vessel. It is ullclear, however, if the residue represents the first food to come into contact with the ceramics, afler which the available binding sites arc saturated , or the las!, if older rcsidues li re continually replaced by new ones, or a combination of all food ever to have been inside the vessel, if the molecules Ihal make up the residue compete for the avai lable binding places. If the fi rst is Ihe ease it must be taken inlo account that many unglazed vessels ure 'scasoned', treated with oil or milk to make them less porous, before bei ng used. This means Ihat the eermnic matrix can indeed be saturated, but also that the residues will tell us more about the seasoning agent than ubout the actual use of Ihe vessel. If we assume, on the other hand, that residues are easily replaccd, we must account fo r the fact that vessels may end up on trash dumps or in graves where they can eomc into contact with organic materials thaI are not related to their ori ginal use. Finall y, if there is competi tion for the available binding placcs it will be difficu lt to separate these diffe rent sources, espec ial ly if vesscls havc bcen in use for a long time for a varicty of purposes. Organic residues in potsherds may also originate from sour? ces other than food but nevcrthcless related to Ihe use of Ihe vessel. Cerami cs Cli n have been employed fo r ' industr ial' purposes (such as the preparation of organic dyes or glues) or as censers, smoking pipes or coffins (especiall y for infanls), or to store a multitude of things. On the other end of the spec? trum are foodstuffs that never come into contact with ceramic vessels but are eaten raw, roasled over a fire or prepared and consumed in other ways that do not involve ceramics. Archae? ological residue analysis, as discussed here, can therefore never be more than complementary in the reconstruction of the dict of an ancient people. For our laboratory tec hniques to reach their full potential as archaeological tools the sludy of more material with experimental or ethno-archaeological origins wi ll prove as indispensable as a funhe r cooperation and shari ng of dala by those working in Ihis field. Acknowledgments Support for stable isolope ratio mass spectrometry instru? mentation at the Envi ronmental Isotope Paleobiogeochemistry Laboratory was provided by the National Science Foundation (USA), grant SBR 98-7 1480. We would like to thank Rene Cappers, Alek Dooley, Jclmer Eerkens, Ernestinc Elster, Richard Evershed, Kym Faull, Mohsen Kamel , Wim Van Neer, Willeke Wcndrich and two anonymous reviewers for thcir help in preparing this article. References [II I.H. Abu-t.eiha, Composi lion of camel milk, Milchwilisenschafl 42 ( t987) 368-371. (2] I. H. Abu?Lciha , I>hysicat and chemical characleri slics of camel milk fat and ilS frac tions, I'ood Chemi~try 34 (1989) 261-272. PI J. Allen, M.E. Newman. M. Riford. G.H. Arther, Btood and planl resi? dues on tlawli ian stonc tool!! from two archacological silcs in upland Kane?ohc. Ko' 011\1 Poko District. O'.hu Island, Asian Penpcctivcs 14 (1995) 283- 302, (4) M.A, Alshaikh. M.S. Salah. Erfeci of milking interval on lIec~t i(m rate lind composition of camel milk in lilte lactation. Journal of Dai ry Re? .? earch 6 1 (1994) 45 t-456. [51 S.H. Ambrose. Isotopic ~na l ysis of paleodiets. Methodological and inter? pretive con~ide rations, in: M.K. Sanarord (Ed.). Invesli!31ion$ of Ancient Human TIuuc. Chemi~al AnalYlics in Anthropology, Gordon and Orclleh Scicnce Publisher~. Langhorne. 1993, PI'. 60- 130. 16) R.W. Bull ieL. The Camel and the Wheel. Harvard Univcr$ily Prc~s. Cambridge, 1975, [7J C. Ca llaneo, K. Gclsthorpe. P. Phillip~. R.J . Sokol. Blood residue~ on stone tools. Indoor and oU ldoor experiments. World Archaeology 25 (1997) 29- 43. (8) DE CraiS. M .l. Coll ins, An improvcd method for 1m: immunotogical de? teelion of miocral bound protein usin, hydronuoric acid and direct cap. ture. Journal of Immunological Melhods 236 (2000) 89- 97. [9J D.E. Craig. MJ. Collins, The removal of prote in from mineral 8urfllees. Implicalions for residue analysis of arthaeological materials. Journal of Arch.cologica l Seicoccs 29 (2002) 1077- 1082. 110] J.D. Connolly, R.A. Hi ll. Tri lerpenoids. natural product rcports. Royal Society of Chemistry 22 (2005) 230- 248. 1111 M.J. DcNiro. S. EpMcin. Mechanisms of carhon isolope fTlictionalion as? sociated Wilh lipid synlhesis. Science 197 ( t977) 26 1- 263. [t2) M, Dorrill, P.H. Whitehead. The species idc lllifiealion of very old human bloodJlailu , FOfel\Sic Science Internat ional lJ (1979) 111-1 16. [13] E.I'. Downs. J.M. Lowenstein. Ident ification of archaeotogical blood pro? leins. A cau tinnary note. Journal of Archaeological Science 22 (1996) 11- 16. (14) J. W. Eerkens. GCfMS analysis and fally ucid ralios of archaeological 1'01? sherd_ from tile western Great Ba.~in of Nonh America. Archaeoilletry 47 (I) (2005) 83- t02. H. Bornord ('I 0/./ Jo"rno/ of Archoe%8i('o/ Sciellce 34 (2007) 28- 37 37 [151 lA. Eisele. 0 ,0 . Fowicr. G. Haynes, R.A. Lewis, Survival and detec tion of blood res idues on Slone lOols. AntiquilY 69 (1995) 36-46. [161 E M. Elamin. CJ. Wilco~. Milk composilion of Majahcim camels. Journal of Dairy Science 75 (1992) 3 155-3157. [17] G.M. EI?Bahuy, Normal contents of Egyptian camel milk. Veterinary Medical Journal 8 (1962) 7- 12. ] 18[ H. Epstein, The Origin of the Dome~tic Animals of Africa, vol. 2, Afri ? cana Publication Corporation. Leipzig, 1971. [19J R.P. Evershed. N. Tuross. Prmeinaccous material from potsherd and as? sociated soils. Jou rn al of Archaeological Sdences 23 (1996) 429- 436. [201 Z. Farah. R. Rellenma ier, 0 , Atkins , Vitamin conte nt of camel milk. In? ternational Journal of Vitamin and NUlrition Research 62 (1992) 30- 33. [21] Z. Farah, M,W. Ruegg, The size distribution of casein micelles in camel milk, Food Microstructure 8 (1989) 211-212. (22) Z. Farah, T. Streiff. M.R. Bachmann. Manufactu re and characterization of camel milk buuer. Milchwissenschan 44 (1989) 412-414. (23] SJ . Fiedel, Blood from stones? Some mcthodological and interpretive prohlems in blood residue analysis, Journal of Archaeological Science 2J ( 1996) 139-147. [24] B.L. Fankhauser. Prolein and lipid analysis of food residues, in: J.G. Hather (Ed.), Tropical Archaeobolany, Applications and New Devel? opments. RoUlledge, London, 1994, pp. 227-250. [25] B.L. Fankhauser, Amino acid analysis of food residues in pottcry. A field and laboratory sludy, Archaeology in Oceania 32 (1997) 131-140. (26) S.O. Gnam, A.M. Shcreha, Composilion of Libyan camcl'.~ milk, Austra? lian 10urnal of Dairy Technology 41 (l9S6) 33-36. (27) A.M. Gorban, O.M. Izreld in. Mincral contenl of camel milk and colos? tru m. Journal of Dairy Research 64 ( 1997) 47 1- 474. (28) A.M. Gorban. O.M. Izzeldin, Fally acids and lipids of camel milk and colostrum, International Journal of r-ood Science and Nutrition 52 (200 1) 283-287. (291 EA. ~Iansel. M.S. Copley, LAS. Madureira, R.P. EI'cl"lihcd. Thermally produced w-{o-alkylphcnyllalkanoic acids provide ev idence for the pro? cessing of marine produces in archacological poltery vessels, Tel rahe? dron Lent ... 45 (2004) 2999-3002. [30] J.R. Hanson, Diterpenoids. Natural Product Reports , Royal SociCly of Chemistry 18 (2001) 88-94. [3t] L,S. Henrikson. R.M, Yohe II , M.E. Newman, M. Drms. Freshwater crus? taceans as an Aboriginal food resourcc in the northern Greal Basin, Jour? nal of California and Grea t Basin Amhropology 20 ( I) ( 1999) 72- 87. (32) D.L, Ingram. L.E. Mounl, Man and Animals in Hot Environmenls. Springer-Verlag, Bcrlin-lleidclbcrg?New York , 1975. [331 Z. Jardali. Contribution 11 I'elude de la composi tion du lai t de dromadaire, Institu t Nat ional Poly technique de Lorraine, Vandoeune?!es-Nancy. 1988 (unpublished thesis). (34) S.S. Kind, R.M. Clccvely, The use of ammoniacal bloodstain ex tracts in ABO groupings, Journal of Forensic Scienccs 15 ( 1969) 131- 134. [H] K.H. Knoess. The camel as a meat 8nd milk animal, World Animal Rcvicw 22 (1977) 39- 44, [36J B. Kooyman. M.E. Newman, II . Ce ri , Vcrifying the rciialJility of blood residue analy~is on archaeological tools. Journal of Archaeological Science 19 (1992) 265- 269. [37] B. Kooyman, M.E. Newman, C. Cluney, M. Lobb, S. Tolman, P. McNeil, L. Hill. Idenlification of horse exploilation by Clovis hunters based on protein an~lysis, American Antiquity 66 (2001) 686- 69 1. [38J M.E. Malainey, R. Przybylski, B.L. Sherriff. The effec ts of thermal and oxidative dcgradation on the fauy acid composition offood planls and an? imals of western Canada. Impl ications for the identification of archaeolog? ical vessel residues, Journal of Archaeological Science 26 ( 1999) 95- 103. [391 M.E. Malainey, R. Przybylski. B.L. Silcrriff, Identifying the fonnercontents of late Precomact Period pouery vessels from western Canada using gas chromatogrdphy, Journal of Archaeological Science 26 ( 1999) 425- 438. [401 M.E. Malainey. R. Przybylski. B.L. Sherri ff. One person?s food. How and why fish avo idancc may affec l lhe sculement and subsi stence pauems of hunter?gatherers, American Antiquity 66 (2001) 141-161 , [4 1] M.A. Mohamed, Camel Milk Composition, Swedish University of Agri? cultural Science. Uppsala, 1990 (unpublished Ihesis). [421 J. Morton, H.P. Schwarcz, Paleodietary implicalions from isotopic anal? ysis of food residues on prehistoric Ontario ceramics, Journal of Archae? ological Science 31 (2004) 503- 517. [43] l .-P. Ramet, T he Technology of Making Cheese from Camel Mi lk (Cnm? du! drom .. dorius) Animal Product ion and Health Paper Number 11 3, UN Food and Agricultural Organization, Romc. 2001. [441 E.A , Reber. R,P, Evcl"lihed, How did Mississippians prepare maize? The application of compound?specific camon isotope analysis 10 absorbed poHery residues from several Mississippi Valley siles, Archaeomctry 46 (2004) 19-33. [451 Royal Canadilln MOIInted Police (1983), Melhods Manual. Serology Sect ion. Ottawa (not paginated). [46] W.N. Sawaya, J,K, Kalil, A. AI?Shalha t. H. AI-Mohamed, Chemic81 composition and nutritional qualily of camel milk, Journal of Food Science 49 (1984) 744- 747. [47[ K, Schmidt?Nielsen. Desert Animals. Adaptation and Environment, O~ford University Pre~s, Oxford. 1964. [48) O.C. Shanks . M. Kornfeld. ]) ,0 . Hawk, Protein analysis of Bugas-Hold? ing tools . New trends in immunological studies, Journal of Archaeolog? ical Science 26 (1998) 11 83-1 19 1. (49) R.T, Wilson, The Camel, Longman, l.ondon?New YOlk, 1984. [50) R. Yagil. Camels and Camel Milk Animal Production and Health Paper Number 26. UN Food and Agricultural Organization. Rome, 1982. [51] R, Yagil, Z. fil zion, Effect of drought eondilions on the qual ity of camel milk. Journal of Dairy Researeh 47 ( 1980) 159-166. [52) R. Vagi], Z. Etzion, Milk yields of camels (Comr/", dfomedurius) in drought areas. Comparalive Biochemistry and Physiology A 67 (1980) 207-209. [53] R, Yagil, V.A, Sod-Moriah , N. Mcyerslc in, Dehydration and camel blood, American Journal of l'bysiology 226 (1974) 298- 308. [54] R,M. Yohe II . M.E. Ncwman, J.5. Schneider, Immunologica l identifica? tion of small-mammal proteins on Aboriginal milling equipment, Amer? ican Antiquity 56 ( 199 1) 659- 666. [55J H. Zhang, J. Yao, D. Zhao, H. Liu, J. Li . M. Guo, Changes in chemical composition of Alu baclria ll camel milk duri ng lactation, Journal of Dai ry Science 8& (2005) 3402- 3410. [561 hnp:/Iwww."rehhasc.orglresiducl(project web?pagcs).