lable at ScienceDirect

Journal of Archaeological Science 69 (2016) 118e129

Contents lists avai

Journal of Archaeological Science

journal homepage: http: / /www.elsevier .com/locate/ jas

Research papers

Chemical analysis of Late Classic Maya polychrome pottery paints andpastes from Central Pet�en, Guatemala

Christina T. Halperin a, *, Ronald L. Bishop b

a Universit�e de Montr�eal, Canadab Smithsonian Institution, United States

a r t i c l e i n f o

Article history:Received 19 October 2015Received in revised form2 March 2016Accepted 7 April 2016

Keywords:Chemical analysisLA-ICP-MSINAAAncient MayaProvincial politicsPolychrome potteryPaint recipesProduction and exchangePolitical-economy

* Corresponding author. D�epartement d'anthropoloPavillon Lionel-Groulx, 3150 Jean-Brillant, Montr�eal Q

E-mail address: christina.halperin@umontreal.ca (

http://dx.doi.org/10.1016/j.jas.2016.04.0070305-4403/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

This paper examines political-economic relationships among Late Classic (ca. 600e900 CE) politicalcenters in the Pet�en Lakes region, Guatemala, through the chemical analysis of red paints and pastes ofpolychrome vessels. Chemical analysis of red paints was conducted using Laser Ablation InductivelyCoupled Plasma Mass Spectrometry (LA-ICP-MS), and chemical analysis of vessel pastes was conductedusing Instrumental Neutron Activation Analysis (INAA). These analyses indicate that a number of politicalcenters along the western shore of Lake Pet�en Itz�a, namely Motul de San Jos�e, Tayasal, and Flores, hadaccess to different polychrome pottery production communities. Nonetheless, inhabitants of these Pet�enLakes sites moved between, gifted, or exchanged polychrome pottery with each other, indicating thatwestern Pet�en Lakes region centers closely interacted with each other. We suggest that these sites mayhave been part of or allied with the epigraphically known Ik’ polity. As such, we find that one of thestrengths of ‘second-tier’ polities, such as the Ik’ polity, did not depend on an individual site's size ormonumental expression, but on the relationships they forged with other centers.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

This paper examines political-economic relationships amongLate Classic political centers in the western Pet�en Lakes region,Guatemala, through the chemical analysis of red paints and pastesof polychrome vessels. A growing body of hieroglyphic evidencehas revealed that during the Late Classic period (ca. 600e900 CE)small “second-tier” or provincial Maya polities were entangled incomplex webs of political alliances, conflicts, and negotiations withprominent regional capitals across the Southern Maya Lowlands.Since the more dominant primary capitals, such as Tikal, Calakmul,Palenque, and Piedras Negras, were responsible for producing agreater proportion of the known texts, our textual understandingsof political relations between these large regional capitals andsecond-tier capitals are better than those between second-tiercapitals or between these second-tier capitals and other provin-cial centers (Marcus, 1976, 1998; Martin and Grube, 2000). For

gie, Universit�e de Montr�eal,C H3T 1N8, Canada.C.T. Halperin).

example, textual sources indicate that the relatively small-sizedsites of Motul de San Jos�e and Zacpet�en were subordinate to Tikalduring at least part of the Late Classic period (Foias and Emery,2012; Rice and Rice, 2009). These sites are considered to havebeen secondary centers possessing their own provincial domains.

While political provinces are often perceived as the strongestand most stable of political units (LeCount and Yaeger, 2010;Marcus, 1998), textual sources are relatively silent on interactionwithin and between these smaller political centers, such as be-tweenMotul de San Jos�e and other archaeological sites in the Pet�enLakes region of Guatemala. Likewise, little comparative archaeo-logical research has been undertaken between Late Classic centersof similar size in the Pet�en Lakes region, such as between Motul deSan Jos�e and Tayasal. One exception is a ceramic figurine study byHalperin (2014), which indicates that western Pet�en Lakes regionsites (Motul de San Jos�e, Nixtun Ch'ich’, Tayasal, Flores, Trinidad,Buenavista, and Ch€akokot) had political-economic ties with eachother. The current study provides evidence for the movement ofpolychrome pottery and, in some cases the raw materials used toproduce such pottery, among the Pet�en Lakes region centers ofMotul de San Jos�e, Tayasal, Flores, and Zacpet�en, as well as amongPet�en Lakes centers and the primary capital of Tikal. We suggest

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 119

that centers along the western shores of Lake Pet�en Itz�a allied oraffiliated themselves with one another, serving to strengthen pro-vincial relations.

Iron oxides, such as hematite, magnetite, goethite, and lep-odicrocite, are minerals commonly found in red slips and red slip-paints decorating Classic period Maya pottery (Beaudry, 1989;Goodall et al., 2009; Smith, 2013). These slips and slip paintswere applied to the vessel before it was fired at approximately550e1000 �C. Because these minerals are thermally unstable (e.g.magnetite and goethite convert to hematite when heated ca.300e400 �C, lepodicrocite converts to maghemite between 275and 300 �C and maghemite to hematite ca. 700 �C) (Cornell andSchwertmann, 2003, pp. 365e377; Goodall et al., 2009; Walteret al., 2001), mineral analysis of red slips or slip paints can beproblematic for identifying paint recipes. On the other hand,chemical analysis of vessel paints in conjunction with vessel pastescan provide a high-resolution, multi-component perspective ofmaterials composition useful for identifying pottery productionrecipes and sources. While paint and paste composition data mayco-vary with each other to offer a robust designation of productionsource, paint and paste groups may not always co-vary (Backes Jr.et al., 2012; Cecil and Neff, 2006; Vaughn, 2005). Paste recipesand sources may have been shared while paint recipes (or the rawmaterials used to produce them) were not, or paint recipes (or theraw materials used to produce them) were shared while pasterecipes and sources were not. Thus, multiple lines of evidence aregarnered here to provide a more nuanced understanding of thecomposition, trading, and artisan-patron relationships surroundingpolychrome pottery and their links to political-economic relationsamong Late Classic Maya centers.

2. Archaeological and epigraphic context

The Late Classic Maya political landscape comprised multiplepolities with shifting domains and relations of power (Chase andChase, 1996; Foias, 2013; Fox et al., 1996; Martin and Grube,2000; Rice, 2004). These political relations included centralizedpolities with paramount rulers at regional capitals who were incharge of their own provinces and oversaw, to varying degrees, thepolitical and religious affairs of secondary and tertiary polities andtheir provinces. In some cases, annexed polities may have brokenaway from their paramount rulers or co-existed in relative auton-omy alongside such dominant powers (LeCount and Yaeger, 2010;Marcus, 1993). In turn, these smaller second- or third-tier politiesmay have exhibited their own shifting centralized and decentral-ized relationships on smaller provincial scales. One of the keyquestions explored here is whether smaller Late Classic politiesforged confederacies or loose alliances with each other and if so,how might we identify such relationships archaeologically?

Polychrome pottery had important political roles (Callahan,2014; Foias, 2007; Halperin and Foias, 2010; LeCount, 1999; Rice,2009a, 2009b). Elites patronized the production of polychromepottery, and such pots were gifted or exchanged hands betweenelite patrons and subordinates or among different noble families asmaterial expressions of dominance, alliance, and affiliation. Morethan a reflection of elite status and affiliation, however, polychromepots were also serving vessels for chocolate and corn-based drinksas well as a variety of foodstuffs. As such, they were instrumental inthe political obligations and alliances formed through hosting andbeing invited to feasts.

Current scholarship suggests that the archaeological site ofMotul de San Jos�e was the capital or one of several capitals of theLate Classic Ik’ polity, named for its Emblem Glyph (political titlesand designations of rulership) featuring the ik’ (wind) sign (Fig. 1)(Foias and Emery, 2012). It is a relatively small site at approximately

4.2 km2 in size with a ceremonial core approximately 1.4 km2 and alarge Acropolis complex (Group C) that likely served as the site'sprincipal royal court. Some of the richest sources of written evi-dence on the Ik’ polity derives from a corpus of largely unprove-nienced polychrome pottery in the Ik’ style. Ik’ style pottery detailsthe histories and ritual lives of the Ik’ royal family (Just, 2012;Reents-Budet et al., 1994, 2012). Because the Ik’ style vesselspossess multiple sub-styles and derive frommultiple paste recipes,Reents-Budet et al. (1994, 2012) have suggested that the Ik’ politymay have been composed of a confederation of noble families ormultiple political centers who shared or rotated power amongstthemselves (see also Rice, 2009a).

Excavations at the northwestern edge of Motul de San Jos�e’sAcropolis complex provide archaeological evidence that Motul deSan Jos�e elites patronized the production of polychrome pottery.These excavations uncovered large secondary middens containingdebris from a polychrome pottery production workshop or work-shops, including paint pots, possible bone painting tools, iron oxidemineral pigments (goethite and hematite) possibly used for thepreparation or red paints, ceramic burnishing tools, a figurinemold,burnt clay, large quantities of ash, and wasters of elaborate Ik’ stylepolychrome vessels, non-Ik’ style vessels with hieroglyphs andfigural scenes, and simple polychrome vessels with abstract orgeometric designs (Halperin and Foias, 2010; Smith, 2013). Previ-ous INAA analysis of vessel wasters from this context reveals thatthe pastes of these wasters chemically match those of unprove-nienced Ik’ style vessels as well as excavated pottery from the site(Halperin and Foias, 2010; Reents-Budet et al., 2012). While ar-chaeologists have found possible evidence of a polychrome potteryneighborhood at the site of Tikal, direct archaeological evidence ofpottery workshops is rare in the Maya area (Halperin and Foias,2012), leaving current scholarship to turn to the indirect evidenceof paste and paint recipes among other analyses of the finishedproducts (Foias and Bishop, 1997; Rice, 1984, 2009a).

In addition to Motul de San Jos�e, it is possible that the site ofTayasal was part of or allied with the Ik’ polity. Not only is the sitelocated close to Motul de San Jos�e, but an Ik’ style vessel wasexcavated from a child's burial at the site (Fig. 2). A member of theIk’ royal family may have given this vessel to an elite family living atTayasal to foster alliances. Alternatively, Ik’ royalty or their close kinmay have resided at Tayasal. Tayasal's monumental architecturewas similar in size to Motul de San Jos�e as its Late Classic cere-monial core zone (centered at the “Main Group”) was approxi-mately 1 km2 and included a single large Acropolis complex (Chase,1983; Cowgill, 1963). Likewise, both sites possessed Late to Termi-nal Classic carved stone stela monuments, suggesting they wereboth important political seats of power (Barrios, 2010; Morley,1938, pp. 426e429; Tokovinine and Zender, 2012). Our un-derstandings of the relationships between the sites, however,require further investigations.

Likewise, it is unclear whether the smaller sites of Flores, a smallisland site (0.5 km2) adjacent to the site of Tayasal, and Zacpet�en, apeninsular site (0.23 km2) located approximately 20 km to the eastof Motul de San Jos�e, were politically tied to Motul de San Jos�eduring the Late Classic period. Zacpet�en and Flores were importantpolitical centers during the Postclassic period and may have heldprominent roles during earlier times (Rice and Rice, 2009). Less isknown about these sites during the Late Classic period, however,since their settlements are largely covered by Postclassic andcontemporary period ones.

We propose three possible models of political-economic re-lationships among Lake Pet�en region centers, which we examinethrough the chemical analysis of polychrome vessel paints andpastes excavated from the archaeological sites of Motul de San Jos�eand its satellite sites, Tayasal, Flores, Zacpet�en, and Tikal. (1) Motul

Fig. 1. Map of settlement hierarchies of Tikal and sites in the western Pet�en Lakes region (following sizes of triangles) within inset map of Maya area showing location of the Pet�enLakes region (right).

Fig. 2. Rollout photograph of Ik’ style vessel excavated from Tayasal (K2707 © Justin Kerr).

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129120

de San Jos�e, as the capital of the Ik' polity, was the principal hub ofpolychrome pottery production and distribution in the Pet�en Lakesregion, and as such had a tight control over this prestige good. Inthis sense, we would expect to see polychrome pottery fromdifferent Pet�en Lakes centers with the same chemical paint andpaste groups as those fromMotul de San Jos�e. This model reflects amore hierarchical political-economic relationship between Motulde San Jos�e and its neighboring Pet�en Lakes region sites.

In turn, the movement and exchange of polychrome potterybetween Pet�en Lakes region centers and Tikal may provide furtherinsights into Pet�en Lakes political dynamics. If Motul de San Jos�ewas the dominant political capital in the Pet�en Lakes region, wemight expect that Motul de San Jos�e elites disproportionatelyinteracted and exchanged pottery with the more prestigious capitalof Tikal. Thus, Motul de San Jos�e elites may have disproportionatelyreceived polychrome vessels from Tikal.

(2) The second model proposes that Motul de San Jos�e was oneof several Pet�en Lakes sites where polychrome pottery productionoccurred. These sites, however, may have been intimately tied toone another through political alliances and shared exchange sys-tems. Although written texts are silent on relationships between

these sites, some of the ways in which such political ties may haveoccurred was in the gifting of polychrome vessels between elitefamilies from Pet�en Lakes centers, the movement of Ik' royal familymembers from one Pet�en Lake center to another (e.g., throughmarriages), tightly bounded market systems, and the sharing ofknowledge or exchange of resources among Pet�en Lakes sites, suchas through polychrome pottery artisans who at the behest of theirelite patrons may have apprenticed or visited polychrome potteryworkshops at other Pet�en Lakes centers. For example, written ev-idence indicates that some artisans moved to different centers andworked for more than one elite patron (Houston, 2009; Just, 2012).In this sense, each site may exhibit their own paint and paste rec-ipes with some evidence of pottery, raw materials, or knowledgeexchanged between them.

(3) The third model proposes that Pet�en Lakes sites, especiallythose equal in size to one another, were relatively autonomousfrom one another. Their production systems may have co-existed,but little to no evidence of gifting, shared knowledge, or ex-change of raw materials existed between them. Here we wouldexpect to see each site possessing their own paint and paste recipeswith little or no overlap between them. In this case, each site drew

Fig. 3. Late Classic polychrome vessels with geometric or simple designs: (a) MTJ 341; bowl, Saxche-Palmar polychrome, paint group #3 [MSJ]; (b) FRS112, dish, Saxche-Palmarpolychrome, paint group #3 [MSJ]; (c) FRS135, bowl/vase, Saxche-Palmar polychrome, paint group #6; (d) ZPN002, bowl, Saxche-Palmar polychrome, paint group #1 [FRS]; (e)TSL122, bowl/dish, eroded polychrome, paint group #3 [MSJ]; (f) ZPN003, tripod dish, Zacatal Cream polychrome, paint group #1 [FRS]; (g) TSL106, bowl, Saxche-Palmar poly-chrome, paint group #5 [TSL].

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 121

on their own local sources of clay and pigments and the knowledgeto produce them or had access to their own pottery workshops orartisans whether they were located locally at the site or elsewhere.

In both the second and third scenarios, elites from Pet�en Lakescenters may have interacted directly with elite patrons, merchants,or artisans from Tikal without the mediation or intervention ofMotul de San Jos�e elites. Here we would expect to see relativelysimilar frequencies of Tikal imports at Pet�en Lakes centers.

3. Materials and methods

Polychrome pottery analyses consisted of LA-ICP-MS on 143 red

paint samples and INAA on 370 paste samples (Appendices 1, 2, & 3in supplementary documents). The red paint samples derived frompolychrome vessels were excavated from Motul de San Jos�e,Tayasal, Flores, Zacpet�en, and Tikal (Figs. 3 and 4). The samplesincluded both polychrome vessels with hieroglyphic texts andfigural scenes (including Ik’ style vessels) as well as polychromevessels with geometric or abstract designs. These latter vesselsbelonged to Zacatal Cream and Saxche-Palmar polychrome typesdesignated in Maya typology systems (Smith and Gifford, 1966).The samples submitted for paint analyses were dictated by theirsize, as whole or large partial vessels could not be exported fromGuatemala to the U.S. As such, these samples derive from primary

Fig. 4. Late Classic Polychrome vessels with hieroglyphic texts and/or figural scenes: (a) MTJ252, vase, Ik’ style, fragment of rim text [left] with black scalloped interior rim [right],paint group #3 [MSJ]; (b) MTJ320, vase, Zacatal Cream polychrome, ruler's hand extended in front of cloth tribute bundle with scribe or artisan seated below pink and red hi-eroglyphic bench and holding stylus, paint group #3 [MSJ]; (c) MTJ055, vase, Zacatal Cream polychrome, fragment of rim text, paint group #6 [TIK]; (d) TIK113 dish/plate, Saxche-Palmar, lower torso of dancing deity wearing jaguar attire, paint group #6; (e) FRS104, vase, Zacatal Cream polychrome, fragment of rim text, paint group #1 [FRS]; (f) TIK099, vase,Zacatal Cream polychrome, detail of feathers, paint group #4 [TIK]; (g) FRS124, plate/dish, Zacatal Cream polychrome, seated elderly deity, paint group #1 [FRS]. (For interpretationof the references to color in this figure legend, the reader is referred to the web version of this article.)

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129122

and secondary midden contexts rather than caches or burials. Asidefrom this consideration and a preference for including Ik' stylesherds and pottery from the production midden at Motul de SanJos�e, all other samples were chosen randomly.

Chemical analysis of the vessel paints was conducted by Hal-perin using LA-ICP-MS at the Field Museum of Natural History inChicago. LA-ICP-MS was conducted with an Analytik Jena ICP-MScoupled with a New Wave Research UP213 laser system followingthe facility's protocols (Dussubieux et al., 2007). The New WaveUP213 laser operates at a wavelength of 213 nm. It operates at 70%

of its maximum energy (0.2 mJ) and at a pulse frequency of 15 Hz inorder to reach the stability and sensitivity requirements of the lasersignal. Helium is used as a gas carrier at a flow rate of 0.50 l/min.Laser ablation is relatively non-destructive since extracted areas arebarely visible to the naked eye. Unlike bulk analysis techniques, itcan target especially thin layers, such as paint applications. A pre-ablation pass line was conducted to eliminate surface contamina-tion before extracting the sample for ionization and injection intothe quadrupole chamber. Laser settings were adjusted to target thepaint and not pass through into the matrix.

Fig. 5. Two-dimensional plot of PCA1 and PCA2 of red paint samples from Late Classic polychrome vessels. Ninety percent confidence interval ellipses designate principal red paintgroups.

Table 1LA-ICP-MS chemical paint groups from Late Classic polychrome vessels (columns) and archaeological provenience (rows).

Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 Unassigned Total

Flores Flores Motul de San Jose Tikal Tayasal Unknown

Flores 45 10 2 0 0 2 1 60Motul de San Jos�e 0 0 34 2 2 1 2 41Tikal 0 0 0 7 0 2 4 13Tayasal 0 0 4 0 16 0 1 21Zacpet�en 3 1 0 0 0 0 0 4Total 48 11 40 9 18 5 8 139

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 123

An internal standard, the isotope Si29, was used to correct forpossible instrumental drifts or changes in the ablation efficiency.This standard was chosen since the red paints are slip-based paintscontaining silica in high concentrations. Si was monitored to makesure its signal was in the same intensity ranges as previous pasteanalyses with detection limits between less than 1 ppb for someelements and 2 ppm for Cu (for detection limits of each element,see Dussubieux et al., 2007:Fig. 1). Two National Institute forStandards and Technology (NIST) standard reference materials(SRM), 679 and 610, were used as external standards to quantify thedata. SRM 679 (Brick Clay) is a clay powder that is shaped into apellet and fired at 900� for an hour and is ideal for ceramic-basedmaterials. SRM 610 was used to calculate the trace element con-centrations. The methods for calculating concentrations followedthose proposed by Gratuze (1999). Ohio Red clay, which is similar tothe matrices of the paint samples, was also analyzed several times aday as a control to ensure the precision and accuracy of the readings(see Appendix 4 in supplementary documents).

Chemical paint groups were identified first by using a hierar-chical cluster analysis to sort samples into preliminary groups. Fouroutliers of the 143 samples were removed. These outliers werelikely due to operational errors as one sample chemically matchedone of the standards and the others were likely due to extremelythin and eroded paint samples where the ablation may havepenetrated past the paint into the paste matrix. All elementalcompositions of the remaining 139 samples were then converted toFe ratios. Since Fe was one of the most abundant elements of redpaint, Fe ratio conversions were made to eliminate its diluting ef-fect and to better identify the chemical variation inherent in a giveniron oxide source (Backes Jr. et al., 2012; Popelka-Filcoff et al.,2007). Bivariate plots of all the element/Fe ratios were thenexamined to identify themost diagnostic elements (Li, B, Na,Mg, Al,Si, K, Ca, Mn, Co, Ni, Sr, Mo, Cs, Ba, Nd, Sm, Eu, Gd, Ta) (Michelaki and

Hancock, 2011). Chemical paint groups were identified based onthe initial hierarchical cluster analysis in combination with Prin-cipal Component Analysis (PCA) of element/Fe ratios using the 20diagnostic elements. Component extraction proceeded from amatrix of elemental covariances. PCAwas run on both logged (base10) and unlogged ratios with no significant difference betweenthem. Group belonging occurred at 90% confidence levels.

Chemical analysis of the vessel pastes was carried out by Bishopusing the Smithsonian Institution's INAA facility at the NationalInstitute for Standards and Technology following routine protocols(see Blackman and Bishop, 2007) with concentrations determinedthrough the use of the comparator standard SRM 1633, coal fly ash(Appendix 5 in supplementary documents). Not all elementalconcentrations were used in the data modeling discussed below.The alkali elements and calciumwere excluded due to their variableproperties of depletion or enrichment in the tropical environment.Other elements were excluded because of a high number of missingdeterminations or issues of analytical precision.

The paste samples overlapwith the polychrome pottery sampleschemically analyzed for paint but also include pottery excavatedfrom Motul de San Jos�e’s satellite centers, such as Buenavista,Ch€akokot, Trinidad de Nosotros (hereafter referred to as Trinidad),and Xilil. These satellite settlements are considered to have beensubordinate centers to Motul de San Jos�e due to their smaller set-tlement sizes and diminutive monumental expressions (Foias andEmery, 2012). Several paste groups can be defined using thechemical characteristics of ceramic pastes within an initial 1400sample set assembled for the greater Motul de San Jos�e and Tikalregions. Analysis was guided by the interest in seeing if there wasevidence for similarity of grouping behavior for the ceramic pastesthat might co-vary with the patterns observed for their paintsamples.

Several statistically refined groups and smaller, distinct but low

Table 2Descriptive statistics of Late Classic polychrome vessel LA-ICP-MS chemical paintgroups (expressed as Fe ratios).

Group 1 Flores (n ¼ 48) Group 2 Flores (n ¼ 11)

Mean SD Min Max Mean SD Min Max

Si 10.50 0.17 10.11 10.85 Si 10.47 0.10 10.32 10.61Na 8.27 0.26 7.75 8.93 Na 8.08 0.25 7.62 8.51Mg 8.80 0.28 8.38 9.99 Mg 8.59 0.17 8.37 9.02Al 9.87 0.16 9.60 10.18 Al 9.77 0.11 9.58 9.94K 8.86 0.21 8.39 9.34 K 8.77 0.13 8.57 8.97Ca 9.29 0.22 8.84 10.09 Ca 9.18 0.13 8.95 9.37Mn 7.77 0.38 7.19 8.99 Mn 7.26 0.13 7.07 7.48Li 6.22 0.25 5.74 6.96 Li 6.06 0.13 5.90 6.27B 6.51 0.21 6.03 7.13 B 6.43 0.14 6.18 6.70Ni 6.60 0.14 6.34 6.97 Ni 6.41 0.09 6.29 6.57Co 6.10 0.30 5.65 7.25 Co 5.72 0.10 5.59 5.94Sr 7.11 0.24 6.67 7.70 Sr 6.93 0.21 6.67 7.24Cs 5.80 0.29 4.76 6.52 Cs 5.66 0.25 5.07 5.91Ba 7.89 0.23 7.45 8.48 Ba 7.69 0.24 7.42 8.10Ta 5.18 0.16 4.88 5.50 Ta 5.20 0.08 5.09 5.33Mo 5.52 0.25 5.12 6.35 Mo 6.23 0.32 5.46 6.75Nd 6.33 0.26 5.91 6.98 Nd 6.04 0.11 5.87 6.18Sm 5.66 0.23 5.24 6.28 Sm 5.36 0.12 5.11 5.52Eu 5.01 0.22 4.61 5.59 Eu 4.70 0.16 4.48 4.93Gd 5.58 0.21 5.16 6.17 Gd 5.31 0.13 5.12 5.44

Group 3 Motul de San Jos�e(n ¼ 40)

Group 4 Tikal (n ¼ 9)

Mean SD Min Max Mean SD Min Max

Si 10.35 0.15 10.06 10.70 Si 10.43 0.11 10.28 10.58Na 7.63 0.22 7.18 8.18 Na 8.86 0.31 8.21 9.26Mg 8.48 0.29 7.94 9.40 Mg 8.60 0.22 8.33 8.93Al 9.75 0.16 9.42 10.16 Al 9.82 0.11 9.61 10.00K 8.18 0.23 7.80 8.65 K 9.05 0.22 8.60 9.34Ca 9.16 0.20 8.74 9.54 Ca 9.40 0.18 9.25 9.80Mn 7.62 0.37 6.67 8.36 Mn 8.13 0.48 7.67 9.08Li 6.47 0.36 5.44 7.05 Li 6.30 0.16 6.03 6.46B 6.48 0.17 6.06 6.78 B 7.04 0.29 6.57 7.46Ni 6.61 0.18 6.20 7.10 Ni 6.75 0.13 6.55 6.90Co 5.97 0.29 5.40 6.53 Co 6.32 0.37 5.96 7.21Sr 6.42 0.17 6.03 6.88 Sr 6.68 0.16 6.46 6.88Cs 5.89 0.29 5.38 6.41 Cs 5.48 0.19 5.26 5.84Ba 7.58 0.16 7.20 7.96 Ba 7.81 0.51 7.16 8.50Ta 5.12 0.16 4.78 5.38 Ta 5.07 0.12 4.89 5.23Mo 5.28 0.24 4.99 6.26 Mo 5.33 0.10 5.21 5.48Nd 6.04 0.16 5.68 6.29 Nd 6.26 0.22 5.89 6.55Sm 5.37 0.16 5.01 5.66 Sm 5.59 0.22 5.21 5.83Eu 4.71 0.15 4.34 4.99 Eu 4.93 0.25 4.49 5.20Gd 5.29 0.15 4.96 5.57 Gd 5.52 0.22 5.15 5.76

Group 5 Tayasal (n ¼ 18) Group 6 Unknown (n ¼ 5)

Mean SD Min Max Mean SD Min Max

Si 10.44 0.15 10.09 10.66 Si 10.47 0.13 10.35 10.61Na 7.56 0.26 7.02 7.97 Na 8.62 0.31 8.30 9.07Mg 8.44 0.19 8.15 8.74 Mg 8.86 0.21 8.53 9.05Al 9.97 0.14 9.75 10.22 Al 9.93 0.10 9.80 10.04K 7.91 0.26 7.56 8.51 K 8.90 0.12 8.77 9.07Ca 9.13 0.18 8.82 9.40 Ca 9.39 0.14 9.24 9.54Mn 7.80 0.57 6.97 8.78 Mn 8.25 0.44 7.64 8.79Li 6.87 0.26 6.34 7.25 Li 6.47 0.33 6.17 7.02B 6.58 0.16 6.30 7.00 B 6.82 0.50 6.24 7.52Ni 6.95 0.21 6.60 7.30 Ni 6.84 0.22 6.60 7.05Co 6.05 0.36 5.55 6.88 Co 6.44 0.23 6.19 6.68Sr 6.40 0.18 6.13 6.79 Sr 6.95 0.15 6.68 7.04Cs 5.73 0.41 5.14 6.53 Cs 5.68 0.14 5.48 5.82Ba 7.71 0.21 7.48 8.23 Ba 7.92 0.21 7.79 8.29Ta 5.31 0.19 5.02 5.68 Ta 5.20 0.18 4.98 5.40Mo 5.26 0.23 4.77 5.59 Mo 5.43 0.34 5.10 5.90Nd 6.32 0.24 5.94 6.79 Nd 6.95 0.48 6.36 7.60Sm 5.67 0.23 5.26 6.02 Sm 6.23 0.36 5.82 6.74Eu 5.00 0.20 4.55 5.32 Eu 5.53 0.35 5.07 6.00Gd 5.59 0.21 5.18 5.94 Gd 6.08 0.34 5.65 6.54

Fig. 6. Bivariate plot of groups formed using elemental concentrations of the ceramicpaste using discriminant analyses 1 and 2. Separation of the paste groups are shownrelative to the first two axes derived from a discriminant analysis. The elementalcorrelations with the axes are given in Table 5. Cluster 5 (represented by the “þ” sign)is shown by its projection onto the axes obtained by the analysis of compositionalgroups 1, 6, 9, 10, 12, and 13. The cluster is similar to group 9 but differs in havingelevated concentrations of several elements, especially the light rare earths; members,however, are too few for statistical refinement. Groups were refined at the 95% con-fidence level but ellipses shown here call attention to the group locations at the 90%confidence level.

Fig. 7. Bivariate plot of groups formed using elemental concentrations of the ceramicpaste using discriminant analyzes axes 3 and 4. The tendency toward separationamong groups 1, 10 and 6, which are overlapped in Fig. 6, are shown using discriminantaxes 3 and 4.

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129124

membership clusters were found in the composition of the paste.

Some of these groups or clusters, such as those pertaining to Pre-classic and Early Classic decorative expression or with obviousdifferences in the texture or temper provide little informationalvalue in the present study. Other reference units were formed and

Table 3Correspondence of select INAA pottery paste groups and clusters (columns) by archaeological provenience (rows). Statistically significant groups indicated by both number andname designations, and clusters indicated by only number designations.

1-Flores 6-Motul 9-Motul 10-Tikal 12-Tayasal 13-Tayasal 5 20 21 23 25 27 30 31

Flores 26 3 3 0 1 0 7 16 0 2 0 0 0 0Motul de San Jos�e 13 19 39 6 2 0 0 0 0 11 1 0 1 12Buenavista (TBV) 1 0 0 0 0 0 0 0 0 0 0 0 0 0Chakokot 0 0 3 0 1 0 0 0 0 0 0 0 0 0Trinidad 8 3 8 1 0 0 1 0 0 0 0 0 0 0Xilil 2 5 0 0 0 0 0 0 0 0 0 0 0 0Tayasal 0 1 1 0 8 31 0 0 0 1 0 8 1 0Zacpet�en 0 1 0 0 0 0 0 3 1 0 0 0 0 0Lake Pet�en 0 0 0 0 0 1 0 0 0 1 0 0 0 0Tikal 2 4 0 44 4 0 0 0 0 0 6 0 1 1

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 125

evaluated but lacked any of the specimens that were the subject ofthe paint analysis (e.g. a well-defined Motul de San Jos�e group thatcontained only figurines). To reduce variation that arises frommanysources and to retain a focus on the extent of Late Classic period co-variation between polychrome pottery surface paint color andpaste composition, we reduced the data set to 307 pottery samples.Thus, the samples for which INAA data are reported here includethe grouped data from the analysis of the ceramic pastes, as well asthe paste concentrations for the samples on which the paint anal-ysis was carried out, whether or not they were members of astatistically-refined paste group. After the removal of outlyingspecimens, the number of samples having both paste and paintdata was 124.

The paste groups were formed through average linkage clusteranalysis using a matrix of Euclidean distances. These groups wereillustrated through the use of discriminant analysis for graphicalpresentation only and not used to evaluate or assign samples to anygroup. The statistical evaluation and refinement of the paste groupswas at the 95% confidence interval about a multivariate groupcentroid using a sample's Mahalanobis distance from that centroid(see Bishop and Neff, 1989).

4. Results

Results of the chemical paint analysis revealed the presence of 6statistically significant chemical paint groups (Fig. 5, Tables 1 and2). These groups clustered by archaeological site, indicating thatmost sites possessed or had access to their own community orclosely linked communities of polychrome pottery producers.Groups 1 and 2 comprised red paint on sherds excavated primarilyfrom Flores. Group 3 comprised red paint on sherds excavatedprimarily from Motul de San Jos�e and included all the sampled Ik'style sherds. Group 4 comprised red paint on sherds excavatedprimarily from Tikal, and Group 5 comprised red paint on sherdsexcavated primarily from Tayasal. Group 6 had only five specimensfrom three different sites. More research is needed to identify apossible source location(s) for this group. A paint group forZacpet�en was not identified as these samples fit in the Floresgroups. The remaining eight samples were unassigned.

In turn, the paste analysis identified several paste groups andsmall, but distinct outlying paste clusters (Figs. 6 and 7, Tables 3e5).Identified paste groups refined at the 95% confidence level con-sisted of Flores (1-Flores), Motul de San Jos�e (6-Motul, 9-Motul),Tayasal (12-Tayasal, 13-Tayasal), and Tikal (10-Tikal). Althoughmore groups than these six were identified, only those of relevancefor the sample also submitted for paint analysis are mentionedhere. The identified clusters may be loosely associated withparticular sites, but the membership of the clusters was small andwas not statistically significant.

Paste samples from Motul de San Jos�e’s smaller satellite sites

(Buenavista, Ch€akokot, Trinidad, and Xilil) were not chemicallydistinguishable from those deriving from Motul de San Jos�e andthus were subsumed under the Motul groups. Such chemical sim-ilarity between these sites is not surprising given their closephysical location to each other as well as the subordinate status ofthe satellite sites. It is possible that some of the smaller satellitecenters produced their own polychrome pottery but shared theirraw materials/finished products with Motul de San Jos�e to such anextent that any chemical distinction between pottery produced atthe capital and its satellite centers could not be detected. Given thescale of Motul de San Jos�e, however, it is more likely that thesmaller sites did not produce their own polychrome pottery andreceived polychrome pottery from Motul de San Jos�e where directevidence of polychrome pottery production exists. Such possibil-ities can be further assessed in the future with the analysis andpublication of the larger corpus of paste samples deriving fromthese satellite sites.

Also not surprising are the paint and paste groups designated toTikal, indicating that it had access to its own polychrome potteryproducing community or communities. Interestingly, paint andpaste group designations were identified on the site level for Floresand Tayasal indicating that these Pet�en Lakes centers, in addition toMotul de San Jos�e, had access to their own polychrome potteryproductionworkshop or workshops. Since little is known about thepolitical-economic relationships between these Pet�en Lakes cen-ters, it gives us some indication of their relative autonomy albeitthrough the lens of polychrome pottery.

Despite the association of chemical paint and paste groups bysite, however, these groups also contained sherds excavated fromother sites, evidence of which represents the gifting, movement, orexchange of rawmaterials or finished vessels between sites (Table 1and 3). These exchanges moved in multiple directions, with poly-chrome pottery from Motul paste groups excavated from Flores,Tayasal, Tikal, and Zacpet�en. Flores paste group pottery was exca-vated from Motul de San Jos�e and its satellite centers as well asTikal. Tikal paste group pottery was excavated from Motul de SanJos�e and one of its satellite centers (Trinidad). Tayasal paste grouppottery was excavated from Flores, Motul de San Jos�e and one of itssatellite centers (Ch€akokot), and Tikal. In turn, Motul de San Jos�epaint group pottery was excavated from Flores and Tayasal, Florespaint group pottery was excavated from Zacpet�en, Tayasal paintgroup pottery was excavated from Motul de San Jos�e, and Tikalpaint group pottery was excavated from Motul de San Jos�e. Theseexchanges included both polychromes with geometric and abstractdesigns and the finest polychrome pottery with figural scenes andhieroglyphic texts.

Most of the samples with assigned paste and paint groupspossessed matching site attributions (69% correspondence; doesn'tinclude paint group 6 of unknown source site) (Fig. 8, Table 6,Appendix 3 in supplementary documents). For example, the Ik’

Table 4Descriptive statistics for groups and cluster formed using INAA chemical concentrations of the ceramic paste.

Group 1-Flores* Cluster 5-Motul Group 6-Motul* Group 9-Motul*

N Min Max Mean CV N Min Max Mean CV N Min Max Mean CV N Min Max Mean CV

Na% 51 0.65 1.21 0.98 13 8 0.83 1.17 1.01 12 35 0.49 1.46 0.83 26 52 0.46 1.11 0.75 21K% 48 0.99 2.27 1.39 20 7 1.24 2.14 1.72 19 35 0.64 2.15 1.38 29 52 0.89 2.40 1.64 23Ca% 49 1.2 4.2 2.1 28 8 1.9 3.2 2.6 17 31 0.9 8.7 1.9 72 46 0.8 4.3 2.0 41Sc used 52 9.4 13.3 10.8 9 8 8.1 9.7 9.1 6 35 10.4 16.0 12.7 11 52 7.4 12.6 10.4 11Cr used 52 25.1 37.9 31.0 10 8 25.3 34.1 28.9 10 35 41.7 81.5 59.2 21 52 22.4 47.1 31.5 13Fe% used 52 2.02 3.66 2.96 12 8 1.99 2.64 2.35 10 35 2.30 3.97 3.15 14 52 1.95 3.04 2.62 10Co** 52 3.37 17.70 6.85 34 8 4.43 11.50 6.37 34 35 3.47 54.60 6.97 121 52 3.01 18.80 7.54 41Zn 52 47 245 72 38 8 52 88 69 19 35 28 134 61 29 52 44 197 72 29Rb 52 45.1 121.0 89.0 19 8 63.5 173.0 131.1 27 35 38.8 147.0 84.2 29 52 61.4 142.0 101.2 18Zr 52 0 381 98 111 8 0 281 72 151 35 0 352 136 87 52 0 331 127 89Sb used 52 1.21 2.20 1.59 12 8 0.68 1.61 1.43 22 35 1.19 1.90 1.54 12 52 1.07 2.40 1.46 16Cs 52 2.56 13.40 5.71 38 8 5.00 6.07 5.56 7 35 2.30 17.50 6.18 63 52 3.42 18.40 7.47 45Ba used 52 611 1870 986 27 8 834 1550 1065 23 35 497 1700 944 31 52 570 1410 901 18La used 52 11.0 20.5 16.6 13 8 31.3 48.0 37.6 14 35 11.6 26.5 17.4 22 52 21.5 52.1 25.9 17Ce used 52 18.5 37.3 27.9 17 8 51.3 70.1 59.3 11 35 20.1 40.0 28.7 17 52 35.7 92.0 48.4 19Sm used 52 1.3 2.9 2.1 18 8 4.1 6.8 5.1 17 35 1.4 5.0 2.9 30 52 3.3 12.1 4.3 28Eu used 52 0.00 0.56 0.42 23 8 0.78 1.27 0.97 16 35 0.25 0.88 0.51 34 52 0.62 2.18 0.78 27Yb used 52 0.96 1.97 1.41 19 8 1.83 3.03 2.26 18 35 0.96 3.71 2.11 29 52 1.97 3.94 2.73 16Lu used 52 0.09 0.37 0.23 25 8 0.30 0.49 0.37 19 35 0.15 0.53 0.31 29 52 0.30 0.65 0.41 17Hf used 52 5.31 8.63 6.50 10 8 5.38 8.57 6.64 18 35 4.85 8.89 6.68 16 52 5.16 10.60 6.61 17Ta** 52 0.66 1.25 0.87 14 8 0.85 1.15 0.95 11 35 0.62 1.13 0.90 11 52 0.75 1.29 0.94 12Th used 52 10.2 16.7 13.5 9 8 13.6 16.0 14.8 5 35 10.9 18.0 13.4 13 52 11.4 16.3 13.3 8U* 52 1.1 5.5 2.6 36 8 2.6 5.3 3.9 23 35 0.0 5.4 2.7 38 52 0.0 4.2 2.4 28

Group 10-Tikal* Group 12-Tayasal 1* Group 13-Tayasal 2*

N Min Max Mean CV N Min Max Mean CV N Min Max Mean CV

Na% 51 0.69 1.79 1.27 17 15 0.62 1.21 0.85 17 32 0.01 0.10 0.04 54K% used 51 1.00 2.44 1.46 19 15 0.92 2.34 1.50 28 2 0.02 0.16 0.09 113Ca% used 45 1.3 7.8 3.1 51 12 0.9 1.9 1.4 21 32 20.0 34.6 26.0 14Sc used 51 8.0 12.2 10.1 11 16 12.8 16.4 14.0 6 32 6.8 10.5 8.5 12Cr 51 27.6 50.1 36.8 14 16 29.2 37.2 32.4 6 32 21.4 36.4 28.4 13Fe% 51 2.22 3.36 2.76 11 16 2.29 3.43 2.61 10 32 1.60 2.86 2.25 14Co** 51 4.10 16.10 7.56 34 16 2.37 11.90 4.62 62 32 3.40 8.00 5.88 18Zn 51 48 86 66 16 16 56 116 85 19 32 27 74 41 23Rb used 51 56.2 132.0 92.7 20 16 58.3 157.0 94.2 31 1 26.9 26.9 26.9 100Zr 51 0 335 110 84 16 0 284 75 154 32 0 0 0 0Sb used 51 0.86 1.65 1.25 12 16 1.21 1.89 1.57 11 32 0.49 1.10 0.84 16Cs used 51 2.88 15.90 5.10 59 16 2.99 9.89 5.21 36 19 0.64 2.83 1.25 57Ba 51 723 3750 1369 46 16 469 1968 810 45 32 109 886 416 41La used 51 16.5 25.0 20.3 12 16 22.4 39.0 29.8 14 32 17.3 28.5 22.2 15Ce used 51 30.9 53.6 39.4 13 16 36.1 70.6 50.5 22 32 27.9 51.1 39.7 18Sm used 51 2.0 3.7 2.7 18 16 4.9 8.1 6.4 15 32 2.1 5.4 4.1 18Eu used 51 0.34 0.73 0.53 21 16 0.75 1.19 0.97 16 32 0.64 1.04 0.82 14Yb used 51 1.18 2.27 1.65 16 16 3.11 4.90 3.85 12 32 1.73 3.00 2.32 15Lu 51 0.16 0.33 0.24 17 16 0.46 0.67 0.58 11 32 0.29 0.53 0.38 16Hf used 51 4.94 7.36 5.89 10 16 7.48 9.62 8.49 8 32 3.40 5.40 4.23 12Ta** 51 0.57 1.07 0.85 11 16 0.85 1.35 1.17 12 32 0.44 0.76 0.60 14Th 51 10.5 15.2 12.8 7 16 13.4 17.4 15.4 8 32 6.3 9.7 7.8 12U* 51 0.0 3.0 1.5 47 16 0.0 5.0 2.9 62 32 0.0 2.9 1.6 51

* Refined at the 95% confidence level; elements used for group formation and refinement are indicated. ** Advisory only, subject to drill bit contamination.

C.T.Halperin,R.L.Bishop

/Journal

ofArchaeological

Science69

(2016)118

e129

126

Table 5Canonical discriminant function loadings of INAA paste groups normalized by within variance. Note: the elements that were selected for the discriminant analysis are thosethat were used to form the chemical paste groups.

Df 1 Df 2 Df 3 Df 4 Df 5 Df 6 Df 7 Df 8 Df 9

Sc 0.254 0.684 0.642 0.754 0.362 0.859 0.075 0.258 0.432Cr �0.098 1.140 �0.287 �0.415 0.251 �0.443 0.124 �0.021 �0.086Fe �0.154 �0.730 �0.324 �1.119 �0.356 0.335 �0.447 �0.401 �0.357Sb �0.185 0.209 0.436 �0.334 �0.291 �0.227 0.441 0.682 0.110Ba �0.077 �0.163 0.283 �0.070 0.448 0.059 �0.428 0.043 0.335La 0.206 �0.210 0.260 �0.474 0.815 �0.223 0.502 �0.243 0.670Ce 0.487 �0.081 �0.283 �0.045 0.138 0.104 �0.443 0.345 �0.375Sm 0.430 0.250 0.238 0.323 0.099 0.253 �0.313 0.038 �1.359Eu �0.102 �0.087 �0.577 0.362 �0.344 0.338 0.696 0.304 0.698Yb 0.065 0.184 0.261 �0.231 �0.403 �0.692 �1.020 0.555 0.306Lu 0.338 �0.063 �0.219 0.070 �0.126 0.084 0.465 �0.944 0.279Hf �0.117 �0.083 0.238 �0.002 �0.351 �0.500 0.190 �0.371 0.026Th 0.062 �0.750 0.366 0.200 0.112 0.070 �0.029 �0.309 �0.502Cumulative proportion of total dispersion

0.590 0.745 0.873 0.959 0.978 0.987 0.995 0.999 1.000

Fig. 8. Bivariate plot showing location of paint groups relative to the groups formedusing the INAA data from the ceramic paste. Groups are shown as in Fig. 6 except that asymbol for paint group has been used; it is superimposed over its location based onpaste chemistry.

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 127

style pottery sampled here belonged only to Motul de San Jos�edesignated paint and paste groups. Likewise, the majority of Floresdesignated paint groups were found on Flores designated pastegroups and so on for each site.

Table 6Cross tabulation of LA-ICP-MS paint groups (columns) by INAA chemical paste groups (r

Group 1 Group 2 Group 3

Flores Flores Motul

1-Flores (group) 16 6 06-Motul (group) 2 1 69-Motul (group) 2 0 710-Tikal (group) 0 0 112-Tayasal 1 (group) 0 1 313-Tayasal 2 (group) 0 0 05-Motul (cluster) 4 2 0Paint group sherds attributed 24 10 17Paint groups sherds not attributed 18 1 22

Nonetheless, some paint groups did not match the same sites asthe paste groups. Such discrepancies may indicate that either (1)potters/painters from these sites were getting clays from other sitesbut using their own local paints or (2) people were exportingvessels to other sites but using the same paint recipes and rawmaterials as those from other sites in the Pet�en Lakes region. Thesecond scenario may be more plausible since ethnographic dataindicate that paint pigments tend to travel farther than clays(Arnold, 1985 Table 2.3; DeBoer and Lathrap, 1979). In addition,Maya hieroglyphic evidence indicates that some artisans may havemoved and worked for different patrons in different political cen-ters (Houston, 2009; Just, 2012). If this was the case, artisans mayhave brought their own pigments with them. As mentioned earlier,such discrepancies in paint and paste recipes also have been foundin other studies of prehistoric pottery production and exchange,such as among the Postclassic Maya in the same Pet�en Lakes region(Cecil and Neff, 2006), among the Gulf Coast Olmec and otherPreclassic sites throughout Mesoamerica (Backes Jr. et al., 2012),and among Nasca sites in the Andes of South America (Vaughn,2005). Even if the paint pigments were more “mobile” betweenpotting communities, both the paint and paste data from the LateClassic Maya polychromes point to the same general pattern: somecenters (Flores, Motul de San Jos�e, Tikal, and Tayasal) had access totheir own polychrome pottery recipes or sources with some evi-dence of the exchange of raw materials or finished products be-tween them.

5. Discussion & conclusions

The chemical paint and paste analyses do not support the firstmodel of a provincial capital centered at Motul de San Jos�e, which

ows).

Group 4 Group 5 Group 6 Unassigned Total

Tikal Tayasal Unknown

0 0 1 0 231 1 0 0 110 1 0 0 101 0 1 0 30 2 0 1 70 1 0 0 10 0 0 0 62 5 2 1 617 12 3 4 67

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129128

served as the primary hub of polychrome production and distri-bution to neighboring centers, such as Tayasal, Flores, andZacpet�en. Rather, Pet�en Lakes sites such as Tayasal, Flores, andMotul de San Jos�e had their own polychrome production work-shops e or at least access to such workshops e as articulated in thesecond or third models. Pet�en Lakes political-elites, however,gifted, moved, or exchanged these vessels with each other, apattern more in line with the second model, in which Motul de SanJos�e was only one among other prominent neighboring centerswhose political influence was expressed in and through themovement of polychrome vessels. Likewise, although it is clear thatMotul de San Jos�e had access to Tikal polychrome pottery imports,several Pet�en Lakes sites (Motul de San Jos�e, Flores, and Tayasal)were tied to Tikal through the direct or indirect exchange of theirpottery with this dominant center. More research is needed toclarify the nature of interactions between Tikal and these otherPet�en Lakes sites.

While Motul de San Jos�e does not appear to have dominated theproduction and distribution of polychrome pottery in the westernPet�en Lakes region, the evidence presented here does not indicatethat Flores, Tayasal, or Zacpet�en produced Ik’ style vessels. TheTayasal excavated Ik’ style vessel and the corpus of unproveniencedIk’ style vessels, however, were not included in the paint analyses.Further paint analyses of thesewhole vessels would be necessary tobetter understand the variability of paint recipes of Ik' style vesselsspecifically. The LA-ICP-MS analyses would require a larger,specialized laser ablation chamber to fit the whole vessels andspecial permission to move the vessels from the individual mu-seums where the vessels are housed. In addition, it is important tokeep in mind that the models do not preclude the possibility thatother types of prestige goods were exchanged (e.g. jade, pyrite,sacred wood objects) between these centers and that political re-lationships may have been forged through other means.

Nonetheless, the polychrome paint and paste analyses pre-sented here support other lines of evidence to suggest that somePet�en Lakes centers may have been allied or shared political powerwith each other. For example, studies by art historians and epig-raphers reveal that Ik’ style vessels depict Ik’ rulers paired side-by-side (Just, 2012; Reents-Budet et al., 2012; Tokovinine, 2013, pp.84e85). These paired Ik’ co-rulers each possessed k'uhul Ik’ ajaw(holy Ik’ lord) royal titles and were paired together within the sameceremonial scene, exhibiting a display of power-sharing relativelyuncommon for Classic period imagery. These visual displays,however, conformwith Contact period evidence of the existence ofjunior-senior pairs of Itz�a rulers from the Pet�en Lakes region (Jones,1998; Rice, 2004). In addition, multiple rulers from closely locatedpolitical capitals, such as from Dos Pilas and Aguateca or fromZapote Babal and El Pajaral, shared the same Emblem Glyphs(Fitzsimmons, 2012; Mathews and Willey, 1991), indicating thatpolitical relationships may have been forged as partnerships inaddition to dominant-subordinate relations.

Thus, we find that any understanding of the Ik' polity mustinclude a regional focus of multiple sites in the Pet�en Lakes regionin addition to its relations with more distant primary politicalcapitals, such as Tikal. Furthermore, we assert that the strength ofthese second-tier or provincial capitals did not necessarily dependon their individual size or the grandeur of their monumental ex-pressions, but on the alliances and networks that they forged witheach other.

Endnote

The comparative samples for the INAA portion of the study weresubmitted by several individuals over the course of several years.The ”Motul area” ceramics were ascribed an analytical number

consisting of “MSJxxx” and have been subjective to previous in-terpretations (e.g., Halperin et al., 2009; Reents-Budet et al., 2012).A more recent shipment of pottery for chemical analysis, unfortu-nately, was given INAA identification numbers that overlappedthose previously assigned. Luckily, it did not compromise earlierinterpretations. This situation has been reconciled with the regis-tration books but required recoding of some of the sample; thus,“MTJ” appears here and in the data listing for INAA samples(Appendix 3).

Acknowledgements

We would like to thank the many archaeological projects fromwhich samples derive: Antonia Foias from the ProyectoArqueol�ogico Motul de San Jos�e, Prudence Rice from the ProyectoMaya Colonial (PMC), Tim Pugh from the Proyecto Arqueol�ogicoTayasal, the late T. Patrick Culbert from the University of Pennsyl-vania Tikal Project, Mario Enrique Zetina Aldana from the RescateArqueol�ogico de Isla de Flores Phase I, and Yovanny Hernandezfrom the Rescate Arqueol�ogico de Isla de Flores Phase II. TheDepartment of Art and Archaeology at Princeton Universitygenerously funded Halperin's LA-ICP-MS analyses at the FieldMuseum of Natural History. The assistance of the reactor operatorat the NIST Center for Neutron Research and Laure Dussubieux fromthe Field Museum of Natural History is greatly appreciated. Hal-perin is grateful to Satish Myneni for his advice on Fe-ratios. We aregrateful to Laure Dussubieux, Antonia Foias, Prudence Rice, MarcosMartinon-Torres, and several anonymous reviewers for theirhelpful comments to the paper. All errors are our own.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jas.2016.04.007.

References

Arnold, D.E., 1985. Ceramic Theory and Cultural Process. Cambridge UniversityPress, Cambridge.

Backes Jr., C.J., Cheetham, D., Neff, H., 2012. The color of influence: a provenancestudy of hematite-based paints on early Olmec carved pottery. Lat. Am. Antiq.23, 70e92.

Barrios, E., 2010. Los Monumentos Prehisp�anicos de Tayasal. Rev. Pet�en Itz�a 73,24e27.

Beaudry, M.P., 1989. Red painted surfaces on selected mesoamerican ceramics: acase study in the use of materials analysis on archaeological ceramics. In:Henderson, J. (Ed.), Scientific Analysis in Archaeology and its Interpretation.Oxford University Committee for Archaeology, Institute of Archaeology, Oxford,pp. 82e111.

Bishop, R.L., Neff, H., 1989. Compositional data analysis in archaeology. In: Olin, R.O.(Ed.), Archaeological Chemistry IV, Advances in Chemistry Series 220. AmericanChemical Society, Washington D.C, pp. 576e586.

Blackman, M.J., Bishop, R.L., 2007. The smithsonian-NIST partnership: the applica-tion of instrumental neutron activation analysis to Archaeology. Archaeometry49, 321e341.

Callaghan, M.G., 2014. Maya polychrome vessels as inalienable possessions.Archaeol. Pap. Am. Anthropol. Assoc. 23 (1), 112e127.

Cecil, L.G., Neff, H., 2006. Postclassic Maya slips and paints and their relationship tosocio-political groups in El Peten, Guatemala. J. Archaeol. Sci. 33, 1482e1491.

Chase, A.F., 1983. A Contextual Consideration of the Tayasal-Paxcaman Zone, ElPeten, Guatemala (Ph.D.). University of Pennsylvania, Philadelphia.

Chase, A.F., Chase, D.Z., 1996. More than kin and king: centralized political orga-nization among the late classic Maya. Curr. Anthropol. 37, 803e810.

Cornell, R.M., Schwertmann, H.C.U., 2003. The Iron Oxides: Structure, Properties,Reactions, Occurrences and Uses, second ed. Wiley-VCH, Weinheim.

Cowgill, G.L., 1963. Postclassic Period Culture in the Vicinity of Flores, Peten,Guatemala (Unpublished Ph.D. dissertation). Harvard University, Cambridge.

DeBoer, W.R., Lathrap, D.W., 1979. The making and breaking of Shipibo-Coniboceramics. In: Kramer, C. (Ed.), Ethnoarchaeology: Impliations of Ethnographyfor Archaeology. Columbia University Press, New York, pp. 102e138.

Dussubieux, L., Golitko, M., Williams, P.R., Speakman, R.J., 2007. Laser ablation-inductively couple plasma-mass Spectrometry analysis applied to the charac-terization of Peruvian Wari ceramics. In: Glascock, M.D., Speakman, R.J.,

C.T. Halperin, R.L. Bishop / Journal of Archaeological Science 69 (2016) 118e129 129

Popelka-Filcoff, R.S. (Eds.), Archaeological Chemistry: Analytical Techniques andArchaeological Interpretation. American Chemical Society, Washington, D.C,pp. 349e363.

Fitzsimmons, J.L., 2012. Janaab’ Ti’ O’ and the transformation of Zapote Bobal,Guatemala. Res. Rep. Anc. Maya Writ. 63, 1e21.

Foias, A.E., 2013. Ancient Maya Political Dynamics. University Press of Florida,Gainesville, FL.

Foias, A.E., 2007. Ritual, politics, and pottery economies in the Classic Mayasouthern lowlands. In: Wells, E.C., Davis-Salazar, K.L. (Eds.), MesoamericanRitual Economy: Archaeological and Ethnological Perspectives. University Pressof Colorado, Boulder, pp. 167e194.

Foias, A.E., Bishop, R.L., 1997. Changing ceramic production and exchange in thePetexbatun region, Guatemala: reconsidering the Classic Maya collapse. Anc.Mesoam. 8, 275e291.

Foias, A.E., Emery, K.F. (Eds.), 2012. Motul de San Jos�e: Politics, History and Economyin a Classic Maya Polity. University Press of Florida, Gainesville.

Fox, J.W., Cook, G.W., Chase, A.F., Chase, D.Z., 1996. Questions of political and eco-nomic integration: segmentary versus centralized states among the ancientMaya. Curr. Anthropol. 37, 795e801.

Gratuze, B., 1999. Obsidian characterization by laser ablation ICP-MS and itsapplication to prehistoric trade in the Mediterranean and the Near East: sourcesand distribution of Obsidian within the Aegean and Anatolia. J. Archaeol. Sci. 26,869e881. http://dx.doi.org/10.1006/jasc.1999.0459.

Goodall, R.A., Hall, J., Viel, R., Fredericks, P.M., 2009. A spectroscopic investigation ofpigment and ceramic samples from Copan, Honduras. Archaeometry 51,95e109.

Halperin, C.T., 2014. Circulation as place-making: late classic polities and portableobjects. Am. Anthropol. 116, 110e129.

Halperin, C.T., Bishop, R.L., Spensley, E., Blackman, M.J., 2009. Late classic (A.D. 600-900) Maya market exchange: analysis of figurines from the Motul de San Jos�eRegion, Guatemala. J. Field Archaeol. 43, 457e480.

Halperin, C.T., Foias, A.E., 2012. Motul de San Jos�e Palace pottery production: re-constructions from wasters and debris. In: Foias, A.E., Emery, K.F. (Eds.), Motulde San Jose: Politics, History, and Economy in a Classic Maya Polity. UniversityPress of Florida, Gainsville, pp. 167e193.

Halperin, C.T., Foias, A.E., 2010. Pottery politics: Late Classic Maya palace productionat Motul de San Jos�e, Pet�en, Guatemala. J. Anthropol. Archaeol. 29, 392e411.

Houston, S.D., 2009. A splendid predicament: young men in classic Maya society.Camb. Archaeol. J. 19, 149e178.

Jones, G.D., 1998. The Conquest of the Last Maya Kingdom. Stanford UniversityPress, Stanford.

Just, B., 2012. Dancing into Dreams: Maya Vases of the Ik' Kingdom. PrincetonUniversity Art Museum, Princeton, NJ.

LeCount, L.J., 1999. Polychrome pottery and political strategies in late and terminalclassic lowland Maya society. Lat. Am. Antiq. 10, 239e258.

LeCount, L.J., Yaeger, J., 2010. Classic Maya Provincial Politics: Xunantunich and itsHinterlands. University of Arizona Press, Tuscon.

Marcus, J., 1998. The peaks and valleys of ancient states: an extension of the dy-namic model. In: Feinman, G.M., Marcus, J. (Eds.), Archaic States. School ofAmerican Research, Santa Fe, pp. 59e94.

Marcus, J., 1993. Ancient Maya political organization. In: Sabloff, J.A., Henderson, J.S.(Eds.), Lowland Maya Civilization in the Eighth Century. Dumbarton OaksResearch Library and Collection, pp. 111e183. Washington D.C.

Marcus, J., 1976. Emblem and State in the Classic Maya Lowlands: An EpigraphicApproach to Territorial Organization. Dumbarton Oaks Research Library andCollections, Washington D.C.

Martin, S., Grube, N., 2000. Chronicle of the Maya Kings and Queens: Decipheringthe Dynasties of the Ancient Maya. Thames and Hudson, London.

Mathews, P., Willey, G.R., 1991. Prehistoric polities of the Pasion region: hiero-glyphic texts and their archaeological settings. In: Culbert, T.P. (Ed.), ClassicMaya Political History: Hieroglyphic and Archaeological Evidence. School ofAmerican Research, Santa Fe, pp. 30e71.

Michelaki, K., Hancock, R.G.V., 2011. Chemistry versus data dispersion: is there abetter way to assess and interpret archaeometric data? Archaeometry 53,1259e1279.

Morley, S.G., 1938. The Inscriptions of Peten. Carnegie Institute of Washington,Washington D.C.

Popelka-Filcoff, R.S., Roberston, J.D., Glascock, M.D., Descantes, C., 2007. Traceelement characterization of Ochre from geological sources. J. Radioanal. Nucl.Chem. 272, 17e27.

Reents-Budet, D., Bishop, R.L., MacLeod, B., 1994. Painting styles, workshop loca-tions and pottery production. In: Reents-Budet, D. (Ed.), Painting the MayaUniverse: Royal Ceramics of the Classic Period. Duke University Press, Durham,pp. 164e233.

Reents-Budet, D., Guenter, S., Bishop, R.L., Blackman, 2012. Identity and interaction:ceramic styles and social history of the Ik’ polity. In: Foias, A.E., Emery, K.F.(Eds.), Politics, History and Economy in the Classic Maya Polity of Motul de SanJos�e, Guatemala. University Press of Florida, Gainesville, pp. 67e93.

Rice, P.M., 2009a. Late Classic Maya pottery production: review and synthesis.J. Archaeol. Method Theory 16, 117e156.

Rice, P.M., 2009b. On Classic Maya political economies. J. Anthropol. Archaeol. 28,70e84.

Rice, P.M., 2004. Maya Political Science: Time, Astronomy, and the Cosmos. Uni-versity of Texas Press, Austin.

Rice, P.M., 1984. Pots and Potters: Current Approaches in Ceramic Archaeology.University of CaliforniaRice, P.M.

Rice, P.M., Rice, D.S. (Eds.), 2009. The Kowoj: Identity, Migration, and Geopolitics inLate Postclassic, Pet�en, Guatemala. University Press of Colorado, Boulder.

Smith, E.L., 2013. Unveiling the Mysteries of Maya Color: Characterization of ClassicMaya Ceramic Vessel Pigments. Unpublished BA honors thesis. Princeton Uni-versity, Princeton, NJ.

Smith, R.E., Gifford, J.C., 1966. Maya Ceramic Varieties, Types, and Wares at Uax-actun. Middle American Research Institute. Tulane University, New Orlenes.

Tokovinine, A., 2013. Place and Identity in Classic Maya Narratives. Dumbarton OaksResearch Library & Collection, Washington D.C.

Tokovinine, A., Zender, M., 2012. Lords of windy water: the royal court of Motul deSan Jos�e in Classic Maya inscriptions. In: Foias, A.E., Emery, K.F. (Eds.), Politics,History and Economy in the Classic Maya Polity of Motul de San Jos�e. UniversityPress of Florida, Gainesville, pp. 30e66.

Vaughn, K.J., 2005. Crafts and the materialization of chiefly power in Nasca. In:Vaughn, K.J., Ogburn, D., Conlee, C.A. (Eds.), Foundation of Power in the Pre-hispanic Andes, Archaeological Papers of the American Anthropological Asso-ciation, No. 14. Arlington, VA.

Walter, D., Buxbaum, G., Laqua, W., 2001. The mechanism of the thermal trans-formation from goethite to hematite. J. Therm. Anal. Calorim. 63, 733e748.