A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA, MEXICO Angela C. Huster and Michael E. Smith We describe the development of a new chronology for the Postclassic site of Calixtlahuaca, Toluca, Mexico. We identify three ceramic phases using discriminant analysis of decorated and plainware types. These phases are consistent with excavated stratigraphy, as well as a series of 54 radiocarbon dates. We then assign absolute dates to the phases using Bayesian analysis of radiocarbon dates and historical information on the date of site abandonment. The resulting chronology identifies three phases at the site: Dongu (A.D. 1130–1380), Ninupi (A.D. 1380–1450), and Yata (A.D. 1450–1530). We then discuss the local and regional implications of the chronology. More broadly, our results demonstrate the utility of using multiple, complementary methods for developing more precise chronologies. Describimos el desarrollo de una cronología nueva para el sitio Posclásico de Calixtlahuaca, Toluca, México. Identificamos tres grupos cerámicos por medio de la clasificación automática tipo k-means y del análisis discriminante de tipos sencillos y decorados. Los grupos se definieron con base en la semejanza de los tipos cerámicos. Investigamos las relaciones estratigráficas entre los grupos, y nuestro análisis confirmó que hay una secuencia cronológica entre ellos; es decir, los grupos son fases cronológicas. Luego presentamos 54 fechas de radiocarbono procedentes de nuestras excavaciones que también confirman las relaciones estratigráficas entre las fases. Proponemos fechas absolutas para cada fase con base en el análisis bayesiano de las fechas de radiocarbono y datos históricos sobre la fecha de abandono del sitio. La cronología resultante identifica tres fases para el sitio: Dongu (1130–1380 d.C.), Ninupi (1380–1450 d.C.) e Yata (1450–1530 d.C.). Comparamos esta cronología arqueológica con la cronología histórica de Calixtlahuaca y el Valle de Toluca. Finalmente, analizamos las implicaciones locales y regionales de la nueva cronología. Los resultados iluminan el desarrollo cultural en el Valle de Toluca en la época Postclásica. Más ampliamente, nuestros resultados demuestran la utilidad del uso de varios métodos complementarios para el desarrollo de cronologías más precisas. T he questions archaeologists ask are increasingly outstripping the level of detail provided by traditional chronologies. In the case of the Aztec Empire, researchers interested in the process of imperial growth, or its influence on trade, immigration, or quality of life, must be able to differentiate between the periods before and after the Aztec conquest of a study region. The Aztec Empire is a difficult target to identify archaeologically; the Triple Alliance formed in A.D. 1428 and fell to the Spanish in A.D. 1521. As one moves closer to the boundaries of the empire, the interval between the Aztec and Spanish conquests becomes increasingly shorter, falling below the 100- to 200-year span of most archaeological chronologies used in the region. In this paper we describe a new archaeological chronology for the Late Postclassic provincial site of Calixtlahuaca, aimed at investigating the effects of Aztec conquest. The resulting chronology identifies three phases at the site: Dongu (A.D. 1130–1380), Ninupi (A.D. 1380–1450), and Yata (A.D. 1450–1530). Calixtlahuaca Calixtlahuaca is a Postclassic site in the Toluca Valley of central highland Mexico (Figure 1). Ethnohistorical data from native historical traditions and early colonial lawsuits indicate that the site was the dominant political center within the Toluca Valley prior to the conquest of the region Angela C. Huster and Michael E. Smith 䡲 School of Human Evolution and Social Change, Arizona State University, P.O. Box 872402, Tempe, AZ 85287-2402 (ahuster@asu.edu; Michael.E.Smith.2@asu.edu) Latin American Antiquity 26(1), 2015, pp. 3–25 Copyright © 2015 by the Society for American Archaeology DOI: 10.7183/1045-6635.26.1.3 3 4 LATIN AMERICAN ANTIQUITY [Vol. 26, No. 1, 2015 Figure 1. Site map with excavation locations. by the Triple Alliance (Mexica or Aztec Empire) in A.D. 1476. The region remained under Mexica rule until the Spanish conquest in A.D. 1521. Calixtlahuaca is located on a hill known today as Cerro Tenismo and at its maximum size covered approximately 264 hectares. The site lacks the single monumental core characteristic of many Postclassic sites; instead, it features multiple clusters of monumental architecture, scattered from the base to the summit of Cerro Tenismo. The Calixtlahuaca Archaeological Project, based at Arizona State University (ASU), conducted two seasons of fieldwork at the site. An intensive surface survey in 2006 documented the extent of occupation, and excavations in 2007 targeted a combination of domestic structures and terraces. This focus complemented earlier work on the monumental architecture by José García Payón in the 1930s (García 1936, 1979). The ASU project excavated a total of 27 locations Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 5 Figure 2. Typical excavated house. across the site, nine of which exposed significant domestic architecture (Figure 2). Descriptions of the fieldwork can be found in Smith et al. (2009) and Smith et al. (2013). Chronological Background Our work at Calixtlahuaca faced two difficulties in establishing chronological phases. The first was the relative rarity of superimposed construction phases or deeply stratified middens. The excavations revealed a high degree of residential mobility within the site; these patterns are discussed in more detail below. The second chronological problem was the limited nature of previous chronological work in the Toluca Valley. Based on his excavations in the 1930s, García Payón proposed that Calixtlahuaca had been occupied from the Formative period through the Spanish conquest. He described a scheme of four chronological periods based on ceramic types and architectural stages, but subsequent scholars (including us) have been unable to replicate or verify this sequence. For later excavations at Teotenango directed by Román Piña Chan, Tommasi de Magrelli (1978) and Vargas (1975) proposed a three-phase division of the Postclassic ceramics from that site. They failed to provide any evidence (e.g., stratigraphy, ceramic counts, radiocarbon dates) supporting their sequence, and it has not been replicated or verified by us or by others. All previous attempts at establishing ceramic chronologies in the region include a high degree of type continuity throughout the Postclassic, suggesting that relative type frequencies would be more useful than phases based on their presence or absence. Based on Smith’s 2002 study of the whole ceramic vessels excavated at Calixtlahuaca by García Payón (Smith et al. 2003), it seemed clear that the major occupation of Calixtlahuaca dated to the Middle and Late Postclassic periods. The site lacks Early Postclassic tradewares such as Mazapan red-on-buff, which is found widely across the Toluca Valley (Sugiura 1998). In addition, it lacks the distinctive orange-on-white poly- 6 LATIN AMERICAN ANTIQUITY chromes found at the Early Postclassic site of Huamango north of the Toluca Valley (Piña 1981), or the types assigned to the Late Classic and Epiclassic phases of the Teotenango chronology (Vargas 1975). The imported ceramics at the site are primarily Late Postclassic or Middle-andLate Postclassic types from the Basin of Mexico, including Aztec III and III/IV black-on-orange, the Late Aztec variants of Guinda (polished redware) black-on-red, and Texcoco molded sahumadores (frying-pan incense burners) (Hodge and Minc 1991). The only indication of a non-Postclassic occupation at the site was the recovery of a few Classic-period Teotihuacán-style vessels and a group of Xoo-phase (Epiclassic) Oaxacan vessels from monumental contexts (Smith and Lind 2005). González and colleagues published two thermoluminescence dates of ceramics from unknown proveniences within the site (González et al. 1999). These dates are A.D. 870 ± 80 and 1520 ± 90. The description of sherd selection suggests that the researchers tried to use sherds dating from different time periods, thereby increasing the probability that their samples do not represent the mean date of occupation or the date of highest population density at the site. Our approach to chronology building employed two steps. First, we generated a sequence of ceramic phases using quantitative ceramic seriation supplemented by stratigraphic data and ceramic cross-ties. Then we used Bayesian modeling of calibrated radiocarbon dates to assign calendar dates to the ceramic phases. Ceramic Seriation Seriation is the process of chronologically ordering archaeological contexts based on the frequencies of artifact types or attributes. In most cases, the technique is applied to pottery, although other types of artifacts, such as jewelry and grave lots, have also been seriated successfully. When applied to ceramics, seriation may use types, varieties, motifs, or attributes, depending on which units are the most chronologically sensitive. Contexts may be grouped and ordered using a variety of quantitative methods, including simple statistics, principal component analyses and k-means clustering, and specially written seriation algo- [Vol. 26, No. 1, 2015 rithms. These methods are used to create either a continuous sequence of units or a series of groups corresponding to archaeological phases. The sequence is then oriented chronologically based on independent information, such as stratigraphy, chronometric dates, or ceramic cross-ties with other regions. For Postclassic central Mexico, successful ceramic seriations have been carried out with both types and attributes. Smith (1983) seriated Postclassic ceramics from Xochicalco and Coatetelco in Morelos using 25 type counts and 7 ratios of ordinal-scale attributes. Then Smith and Doershuk (1991) applied the resulting ceramic phases to domestic deposits at Cuexcomate and Capilco using discriminant analysis based on 27 ceramic types. At Yautepec, Hare and Smith (1996) selected 14 types that seemed to have chronological significance and used them to seriate domestic deposits, identifying four phases (Middle Postclassic, Late Postclassic A, Late Postclassic B, and Early Colonial). At Chalco, Hodge (2008) employed 60 types and variants to distinguish three phases (Early, Middle, and Late Postclassic). For surface collections from the Basin of Mexico, Garraty (2009) used three attribute categories, each with three to five possible states, to identify five phases (Middle Postclassic through Early Colonial). Because of the lack of prior work in the Toluca Valley, we did not know which types were likely to be chronologically significant. We therefore began with a large number of variables. Only three classes of types were excluded from the seriation: eroded types, non-eroded types with 10 or fewer sherds in the core seriation sample (described in more detail below), and types occurring only in a single context. This resulted in the inclusion of 62 types in the seriation. Table 1 shows mean type frequencies for all types accounting for more than .01 percent of the ceramics of any phase, including some types excluded from the seriation. A few types included separately in the seriation are combined into single entries in the table. The ceramic classification we have devised for Calixtlahuaca defines types based on vessel form, decoration, and paste, generally in that order. The general approach to classification is based on Smith’s work at Aztec-period sites in Morelos. The nature of ceramic categories is Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 7 Table 1. Ceramic Type Frequencies by Phase. Dongu Type In Seri. Aztec imports Aztec III black-on-orange bowls Y Aztec III/IV black-on-orange bowls Y Aztec black-on-red bowls Y Aztec black-on-red pitchers N Coarse orange jars Y Coarse orange basins N Copas Y Texcoco molded/filleted sahumadors Y Texcoco fabric-marked salt vessels Y Aztec orange spinning bowls N Other imports B-5 bowls Y B-11 bowls Y D-6 bowls N G bowls Y Redware bowls, other N Other decorated bowls Y Other decorated jars Y Plainwares Eroded bowls N Plain bowls Y Ring-base bowls Y Eroded jars N Plain jars Y Thick-rim jars Y Comals Y Basins Y Plain pitchers Y Spoons N Local Decorated bowls A-1 bowls Y A-3 bowls Y B-0 bowls Y B-1 bowls Y B-2 bowls Y B-4 bowls Y B-10 bowls N C-0 bowls Y C-1 bowls Y D-0 bowls Y E-0 bowls Y E-1 bowls Y E-3 bowls Y E-4 bowls Y E-5 bowls Y E-6 bowls Y E-7 bowls N E-8 bowls N E-9 bowls Y E-10 bowls Y E-11 bowls Y E-12 bowls Y E-13 bowls Y E-14 bowls N Ninupi Yata Mean St Dev. Mean St Dev. Mean St Dev. .06 .00 .05 .01 .01 .00 .01 .00 .07 .00 .25 .00 .15 .05 .03 .01 .02 .01 .25 .01 .36 .08 .26 .01 .01 .01 .02 .08 .32 .02 .63 .22 .49 .02 .04 .02 .05 .23 .97 .04 .80 .14 .39 .16 .10 .07 .07 .04 .50 .13 .89 .26 .45 .69 .36 .24 .13 .09 .60 .39 .04 .04 .02 .06 .00 .06 .02 .10 .14 .10 .15 .00 .19 .06 .06 .02 .00 .01 .02 .09 .04 .21 .04 .00 .02 .06 .28 .20 .02 .02 .00 .00 .00 .09 .16 .04 .08 .00 .01 .00 .22 .68 3.34 4.20 .01 32.86 39.61 .16 .04 .02 .02 .02 3.80 3.59 .05 24.86 25.87 .39 .11 .05 .06 .05 3.71 4.87 .04 3.80 37.71 .64 .43 .03 .07 .01 4.25 3.46 .21 19.27 2.62 .74 .74 .07 .29 .04 7.18 3.98 .01 41.18 2.63 .66 1.35 .03 .06 .04 5.99 5.78 .01 19.18 2.65 .60 1.71 .06 .14 .15 .28 .06 1.21 .13 .01 .00 .00 .02 .16 .07 1.39 .06 .06 .08 .45 .18 .01 .02 .26 1.44 .52 .05 .03 .03 .64 .15 .89 .40 .03 .01 .01 .06 .37 .23 1.02 .20 .20 .19 .54 .36 .04 .06 .29 1.31 .72 .18 .11 .14 .29 .01 1.79 .04 .07 .01 .01 .02 .28 .05 1.30 .10 .02 .02 .13 .04 .01 .01 .17 1.04 .25 .03 .03 .02 .26 .02 1.12 .09 .20 .03 .04 .07 .34 .14 .82 .21 .06 .05 .23 .06 .05 .04 .22 .67 .33 .12 .06 .05 .35 .03 1.91 .01 .04 .01 .01 .01 .17 .01 1.14 .06 .01 .01 .06 .08 .00 .00 .06 .75 .15 .04 .01 .00 .57 .09 .83 .03 .09 .02 .05 .03 .23 .02 .85 .16 .03 .03 .12 .18 .00 .02 .11 .64 .21 .08 .02 .01 8 LATIN AMERICAN ANTIQUITY [Vol. 26, No. 1, 2015 Table 1 (continued). Ceramic Type Frequencies by Phase. Dongu Ninupi Yata Type In Seri. Mean St Dev. Mean St Dev. Mean St Dev. E-16 bowls N .03 .14 .00 .01 .02 .08 B or C bowl, uncertain N .00 .00 .01 .04 .01 .05 B or E bowl, uncertain Y .79 .84 .97 1.06 1.38 .64 Local decorated jars B-0 jars Y .35 .76 .38 .42 .39 .39 D-0 jars Y .04 .12 .07 .21 .01 .02 E-0 jars Y 1.11 1.02 .78 .59 .38 .41 E-2 jars Y 1.75 1.61 1.22 1.15 .88 1.02 Eroded painted jars N .00 .00 .08 .44 .00 .00 B or E jar, uncertain Y .49 1.05 .63 .72 .50 .46 Local ritual forms Biconical censers Y 1.28 .93 1.52 1.25 1.08 .70 Other censers Y .16 .49 .11 .26 .07 .13 Large braziers Y .02 .06 .13 .39 .05 .12 Scored braziers Y .19 .40 .24 .37 .23 .32 Toluca Valley sahumadors Y .28 .54 .15 .27 .16 .31 Pyriform incised vases N .01 .05 .01 .02 .02 .08 Tlaloc jars N .01 .07 .00 .00 .02 .08 Miniature vessels Y .10 .33 .04 .08 .08 .10 Crude unfinished vessels Y .42 .44 .33 .39 .18 .17 Non-vessel ceramic items Spindle whorls, Small Y .01 .03 .00 .01 .09 .38 Spindle whorls, large Y .06 .18 .12 .59 .13 .38 Figurines Y .09 .32 .17 .58 .43 1.03 Sherd disks Y 1.40 3.38 1.35 4.63 1.99 5.65 Worked sherds N .01 .03 .01 .03 .02 .08 Water-worn sherds Y .00 .02 .02 .05 .01 .02 Rattle instruments N .01 .03 .01 .04 .07 .23 Wind instruments Y .08 .32 .06 .20 .04 .15 Pipes N .01 .04 .00 .01 .02 .04 Balls/Pellets N .02 .08 .02 .06 .04 .15 Cylinders (solid) Y .04 .16 .02 .07 .07 .23 Lip plugs N .00 .00 .00 .03 .02 .08 Chunks N .00 .02 .01 .04 .05 .16 Figurine molds N .00 .01 .03 .20 .02 .08 Adornos N .01 .03 .00 .01 .03 .10 Spikes N .00 .02 .01 .06 .02 .08 Uncertain Eroded, uncertain N 3.54 5.02 5.43 9.42 7.77 8.27 Vessel frag, form uncertain N .03 .07 .02 .04 .08 .16 Non-vessel frag, form uncertain N .05 .15 .08 .20 .08 .14 Hollow support or handle Y .25 .36 .37 .31 .25 .21 Note: Frequencies based on DS-3 sample, with averages and standard deviations calculated based on capas with more than 50 sherds. based on previous work on Toluca Valley museum collections by Smith (Smith 2001; Smith et al. 2003), modified for use with sherds rather than whole vessels. Our classification currently consists of six major form categories (bowls, jars, basins, comales [griddles], other vessels, and nonvessel objects), which are each divided into multiple types. Bowls (including simple, tripod, and grater bowls) and jars include several families of local ceramics, based on broad decorative similarities (e.g., red-on-buff is Family E, which contains about 15 individual types, while polychrome-on-white is Family C, which contains four types; see Figure 3). A full list of the ceramic types used by the Calixtlahuaca Archaeological Project is available as Supplemental Table 1. Post- Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 9 Figure 3. Local and imported ceramic types. Aztec types (top row); B-5 and B-11 incised redwares (second row); local B, C, and E types (bottom two rows). classic ceramics from the Toluca Valley are often referred to as Matlatzinca wares (e.g., Castillo [1991]), but we do not use this term due to its multiple meanings (see discussion below). The Calixtlahuaca project excavations recovered approximately half a million sherds from over 1,000 excavated lots (levels within an excavated grid square). Because of the variability in the integrity of the excavated contexts, a smaller sample was identified to serve as the core of the seriation. We chose to use 168 lots of more than 50 sherds each from contexts directly associated 10 LATIN AMERICAN ANTIQUITY with house architecture, middens, or near-house refuse scatters. These lots contain an average of 721 sherds each, and the sample as a whole contains over 121,000 sherds. These contexts include most of the cases of directly superimposed strata. For the remainder of this article, this sample is referred to as the Core Seriation Sample (CSS). Ceramic type counts for the CSS can be found in Supplemental Table 2. K-Means Cluster Analysis The first method applied to the CSS to form groups was k-means clustering. K-means is a non-hierarchical clustering procedure that attempts to minimize the distance among cases within clusters and to maximize the distance between clusters. Based on other central Mexican Postclassic chronologies, we hypothesized that the Calixtlahuaca ceramics would conform to two, three, or four chronological phases. We therefore ran three cluster analyses, one for each of these hypotheses. Unfortunately, these clusters were not supported contextually. Excavation lots from the clusters were not consistently ordered stratigraphically, and lots within the same stratigraphic layer would frequently be assigned to several different clusters. While chronology is often the single most distinctive dimension of variation in archaeological ceramic assemblages, it seems that temporal order was too obscured by statistical noise in this dataset. This may be a result of the large number of types included in the analysis, many of which were evidently not chronologically sensitive. Because we view stratigraphy as the most fundamental chronological marker, we abandoned this approach. Discriminant Function Analysis We next used imported ceramic types with known chronological associations to divide the ceramics into groups or phases, which we tested against the stratigraphy. Of the imported types present at the site, the Aztec black-on-orange types are the most securely dated. In Morelos, Smith (1983) found that the presence of imported Aztec III black-on-orange is the most consistent marker of the transition from the Middle to Late Postclassic period. The Aztec black-on-orange found at Calixtlahuaca consists almost exclusively of tripod grater bowls (molcajetes). We distinguished [Vol. 26, No. 1, 2015 Hodge and Minc’s (1991) molcajete variants D and E as Aztec III; variants H, M1, and M2 as Aztec III/IV; and variant M4 as Aztec IV (Figure 4). Aztec orangeware sherds without enough decoration for a secure assignment were coded as Aztec III. This is a conservative division, placing only those cases with specific motifs in the Aztec III/IV category and only those with clear naturalistic (and likely Spanish-derived) designs into the Aztec IV category. As a ceramic type, Aztec III appeared in the Basin of Mexico and in Morelos in the mid-1300s and remained in use until the Spanish conquest (Hare and Smith 1996; Nichols and Charlton 1996). The situation for Aztec IV is less clear. The type has traditionally been considered to post-date the Spanish conquest, and this category has often included most of the variants that we separate out as III/IV, as well as those that we coded as Aztec IV. With increasing research, it has become clear that at least some variants of what has traditionally been called Aztec IV were first produced several decades prior to the Spanish conquest and continued to be produced after it as well (Charlton et al. 2005). This transitional group has become known as Aztec III/IV, although researchers do not agree on exactly which variants fall into this category. Based on current knowledge, production of Aztec III/IV probably began around A.D. 1500 (Hodge 1998). Overholtzer (2014) also uses the Aztec black-on-orange types as the basis for her seriation of deposits at Xaltocan. Based on these considerations, we used the distribution of Aztec orangeware types to partition the excavated lots. We divided the CSS into two initial groups based on the presence and absence of Aztec orangewares in each stratigraphic layer. Given the near absence of the Middle Postclassic orangeware types at Calixtlahuaca—Aztec I and Aztec II black-on-orange—we assumed that all Aztec orangeware sherds dated to the Late Postclassic, or possibly early colonial, periods. For purposes of chronology building, we made the heuristic assumption that each identifiable stratigraphic layer—which we called capas (see discussion below)—pertained to one and only one chronological phase. We therefore assigned all of the lots from a single stratigraphic layer to the same group. The two original groups (with and without Aztec orangeware sherds) were tested Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 11 Figure 4. Aztec III, III/IV, and IV black-on-orange types. for internal coherence by running the discriminant analysis and checking for lots that were switched by the functions (based on overall type composition) from their original group (based on orangewares) to the other group. If most of the lots from a single stratigraphic layer were switched to the opposite group, the lots were either reassigned or removed from the sample. Once the two groups were “corrected” in this manner, all other excavated lots with more than 50 sherds were added to the analysis, but not assigned to an a priori group. When used in this manner, discriminant analysis will produce a probability of that lot being a member of each previously defined group. We then hand-examined the results and assigned all of the lots from a single stratigraphic context to the most common discriminant analysis phase assignment. In a few cases, this resulted in the division of what had been identified as a single stratum during excavation into two components. Strata with highly ambivalent results (e.g., lots were assigned to two or more phases in approximately even quantities, or a majority of lots showed approximately even probabilities of belonging to two or more phases) were coded as chronologically mixed. The discriminant analysis and subsequent phase assignment procedures produced stratigraphically coherent and consistent results for two groups, signaling the success of this approach. Based on the presence and absence of Aztec orangewares, the two groups could be interpreted as pertaining to the Middle and Late Postclassic periods. We next attempted to subdivide each group into two finer phases. To do this, we returned to the CSS. We first divided the CSS group containing Aztec orangeware into two subgroups: one based on lots containing only Aztec III and one based on lots containing Aztec III/IV (with or without Aztec III). This produced a total of three base groups. The procedure described above was repeated using these three groups, and it again produced stratigraphically logical results for the three groups. No distinctions were made between variants of Aztec black-on-orange suspected to have been imported from the Basin of Mexico and those suspected to have been produced locally. Out of six households with material dating to the last phase, three have very low frequencies of Aztec orangewares and other imports, 12 LATIN AMERICAN ANTIQUITY meaning that the discriminant analysis is not simply dividing lots based on the types that we used to create the base groups. We then attempted to divide the Middle Postclassic group into two phases, using the presence of various local types via further k-means clustering. Although we tried many combinations of variables, we could not produce stratigraphically coherent results, so we gave up trying to subdivide that group. The discriminant analysis probabilities for all analyzed lots can be found in Supplemental Table 3. The three phases resembled the Postclassic chronology at Yautepec (Hare and Smith 1996) in having a single Middle Postclassic phase and two Late Postclassic phases. At this point, we were satisfied that we had created a reasonable three-phase sequence for the ceramics at Calixtlahuaca. We named the phases Dongu, Ninupi, and Yata, from oldest to youngest. The names of ceramic phases are drawn from the three major non-Nahuatl languages traditionally spoken in the Toluca Valley (García 1999). Dongu is an Otomi term meaning “old house” or “abandoned house” (Hernández and Torquemada 2010:49; Lastra de Suárez 1989:218, 2008:312), an appropriate name for the earliest phase at the site. Ninupi is a Matlatzinca term meaning “wind” or “air” (Basalenque 1975 [1642]:19; Escalante and Hernández 1999:175), a nod to the well-known Ehecatl temple at the site. The third phase name, Yata, is taken from Mazahua and means “to divide” or “to let fall,” often referring to houses (Kiemole 1975:350; Rodríguez 2010). We also identified a small number of contexts with approximately even probabilities of belonging to two adjacent phases, and some contexts representing chronologically mixed deposits. An attempt to include smaller lots, with 30 to 50 sherds, produced inconsistent results, demonstrating that this method of phase assignment is dependent on adequate sample size. Stratigraphy Stratigraphy is the backbone of our chronology. We manipulated the ceramic data in various ways, as described above, to produce clusters of types that have a clear stratigraphic ordering. Once we had defined the three phases, we undertook a detailed analysis of the stratigraphic relationships of the phased ceramic lots. We recorded the phase assignments on schematic depictions of the [Vol. 26, No. 1, 2015 stratigraphy for each excavated unit. We call our basic unit of cultural stratigraphy the “capa”: this is a soil zone that pertains to a particular type of depositional activity. Most capas appear to have chronological integrity. For each capa, we tallied the number of lots assigned to each of the three phases. For most capas at the site, all constituent lots phased to only one ceramic phase. For other capas, we assigned the entire capa to the phase of the greatest number of individual lots, in line with our heuristic assumption that all capas pertained to a single phase. This procedure is illustrated in Figure 5, a Harris matrix of the stratigraphy in Unit 316, an excavation of a house and surrounding area. Each box is a capa or an architectural unit (floors and walls). Capa labels are written in Times font to the right (or below) each box. Architectural features are in boxes drawn with a heavier line. The numbers inside the capa boxes indicate the number of excavated lots in that capa assigned to each of the Dongu, Ninupi, and Yata phases (in that order). The tally for the numerically dominant phase for each capa is indicated in bold typeface. Although this is our most complex and “messiest” stratigraphic excavation, the results show strong—but not perfect—stratigraphic support for the seriation results. Of the 17 capas in Unit 316 with at least one lot included in the seriation, nine have lots from a single phase. For five capas, one phase contains at least twice as many lots as any other, and for one capa the dominant phase has only one more lot than a second phase. Two capas—the plow zone and a stone floor deposit—produced inconclusive results; their two seriated lots were split between two phases. When all of the capas were assigned to their dominant ceramic phase, every single case of a stratigraphic relationship between capas of different ceramic phases (there are eight such cases in Unit 316; see Figure 5) was in the correct relative order. This result, which is replicated in other excavation units with stratigraphy and multiple phases, provides strong confirmation for the chronological integrity of the phases, which were defined from ceramic data only. One capa without stratigraphic relationship to any other unit (Capa 7) is shown floating in the figure; it is placed between the Dongu and Ninupi phases based on its seriated lots. Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 13 Figure 5. Harris matrix with lot phase assignments based on ceramic seriation. There are very few single-phase ceramic types (Table 1). Instead, the phases are characterized by differences in the relative frequencies of types. There is a cluster of Basin of Mexico–style types—late Aztec Guinda redwares, Texcoco fabric-impressed salt vessels, and comals—that increase sharply over time, from almost complete absence in the Dongu phase to low but consistent presence in the Ninupi phase and rapid increase in the Yata phase. Among the non-Aztec types, the D-group (red-on-white), and G-group (negative-on-natural), are low-frequency types that almost completely disappear after the Dongu phase. Although we considered these types to be of local origin for purposes of seriation, the provisional results of our INAA sourcing suggest that some of these earlier types may in fact be imported from the southern or western portions of the Toluca Valley (Stoner and Glascock 2013). The C-group (polychromes-on-white) is one of the few types to peak in Phase 4, although the group is present in all phases. There is a general trend toward a higher bowl-to-jar ratio and an increased diversity of vessel forms over time. Radiocarbon Dating We obtained 54 radiocarbon dates from our excavations at Calixtlahuaca. This operation had two goals: first, to test the integrity of the phases based on ceramic seriation and attach them to absolute dates; and second, to date terrace deposits that contained insufficient numbers of sherds to include in the seriation or to otherwise date with ceramics. All samples were processed by the University of Arizona Accelerator Mass Spectrometry Lab. The first 20 dates were processed in 2006 as sample numbers AA78218– AA78237, and the latter 34 dates were processed in 2011 as sample numbers AA98717–AA98749. The samples included 21 high-precision dates. All of the samples consisted of carbonized or partially carbonized wood or other plant matter. The samples used in the chronology research described here were selected from household contexts with good ceramic seriation phase assignments (Table 2). A preliminary discussion of the site chronology, based on the first set of 20 dates, was published in Smith et al. 2013. The present 14 LATIN AMERICAN ANTIQUITY [Vol. 26, No. 1, 2015 Table 2. Radiocarbon Dates. Field no. R-1 Lab no. AA78218 Radiocarbon Age (B.P.) 531 ± 31 δ13C % -24.1 R-2 AA78219 420 ± 21 -24.2 R-3 R-4 AA78220 AA78221 584 ± 38 531 ± 21 -23.5 -23.8 R-5 AA78222 486 ± 36 -23.7 R-6 AA78223 410 ± 36 -24.5 R-7 AA78224 369 ± 37 -23.5 R-8 AA78225 366 ± 38 -24.3 R-9 R-10 R-11 AA78226 AA78227 AA78228 427 ± 21 733 ± 21 743 ± 65 -25.1 -25 -26.8 R-12 AA78229 633 ± 24 -25.7 R-13 AA78230 523 ± 21 -10.3 R-14 AA78231 495 ± 36 -23.8 R-15 AA78232 432 ± 36 -11.1 R-16 AA78233 695 ± 36 -23.8 R-17 R-18 AA78234 AA78235 916 ± 49 769 ± 51 -23.2 -18.5 R-19 AA78236 689 ± 37 -14.2 R-20 R-21 R-22 AA78237 AA98717 AA98718 954 ± 38 817 ± 35 404 ± 20 -24.9 -17.1 -24 R-23 R-24 AA98719 AA98720 455 ± 20 817 ± 66 -25.6 -12.5 R-25 AA98721 870 ± 35 -22.1 R-26 AA98722 574 ± 34 -22 R-27 R-29 R-30 R-32 R-33 R-35 AA98723 AA98725 AA98726 AA98728 AA98729 AA98731 912 ± 35 810 ± 35 909 ± 35 618 ± 35 1596 ± 26 118 ± 34 -24.6 -23.4 -12.1 -27.1 -24.5 -24.1 R-36 AA98732 649 ± 35 -25 Interval (A.D. unless specified) 1314–1357 1387–1443 1435–1489 1604–1607 1296–1420 1327–1343 1394–1435 1329–1340 1396–1461 1429–1523 1572–1629 1446–1530 1538–1635 1447–1531 1538–1635 1431–1485 1253–1293 1157–1327 1343–1395 1287–1329 1341–1396 1329–1340 1396–1439 1327–1343 1394–1454 1415–1517 1595–1619 1258–1320 1351–1391 1025–1214 1160–1299 1371–1379 1262–1320 1350–1391 1015–1172 1163–1271 1440–1500 1601–1615 1420–1455 1042–1107 1117–1285 1044–1101 1119–1254 1299–1369 1381–1424 1031–1208 1167–1274 1034–1209 1290–1404 412–539 1678–1765 1773–1777 1800–1940 1279–1329 1340–1396 Probability (%) 27.1 68.3 94.7 .7 95.4 9.1 86.3 1.7 93.7 76.5 18.9 50.7 44.7 49.0 46.4 95.4 95.4 83.6 11.8 39.3 56.1 3.5 91.9 4.0 91.4 88.5 6.9 67.7 27.7 95.4 94.6 .8 63.1 32.3 95.4 95.4 88.9 6.5 95.4 13.3 82.1 20.7 74.7 60.1 35.3 95.4 95.4 95.4 95.4 95.4 32.7 .7 62.0 43.8 51.6 Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 15 Table 2 (continued). Radiocarbon Dates. Field no. R-37 R-38 R-39 R-40 Lab no. AA98733 AA98734 AA98735 AA98736 Radiocarbon Age (B.P.) 1160 ± 35 454 ± 34 765 ± 35 438 ± 34 δ13C % -25 -24.7 -23.1 -9.7 R-42 AA98738 1925 ± 36 -19.8 R-43 R-44 AA98739 AA98740 600 ± 35 998 ± 21 -25.9 -25 R-45 AA98741 2352 ± 58 -20.1 R-46 R-47 AA98742 AA98743 618 ± 36 870 ± 35 -9.3 -23.7 R-48 R-49 AA98744 AA98745 618 ± 36 433 ± 35 -25.9 -24.5 R-50 AA98746 653 ± 35 -25 R-51 AA98747 390 ± 36 -26.2 R-52 AA98748 368 ± 24 -26.3 R-53 R-54 R-55 AA98749 AA98750 AA98751 351 ± 35 635± 36 436± 35 -10.9 -26 -23.6 R-56 AA98752 366± 35 -25.8 paper supersedes that chronological discussion. Nearly all of the calibrated radiocarbon dates fall into the Middle and Late Postclassic periods (Figure 6). All dates were calibrated using OxCal 4.2.2 (Bronk Ramsey 2009) with atmospheric data from Reimer et al. (2009). Excluding one contaminated modern date (R-28), the ranges on the latest dates extend only a short distance into the Early Colonial period. Four dates (R-33, 37, 42, and 45) have a Formative- or Classic-period age. All of these were from layers underlying cultural strata or from gully fill. These dates most likely Interval (A.D. unless specified) 777–971 1410–1487 1211–1289 1415–1511 1601–1616 37–29 B.C. 23–11 B.C. 2 B.C.–140 A.D. 151–170 195–209 1295–1410 990–1045 1095–1120 1141–1148 751–686 B.C. 668–638 B.C. 621–615 B.C. 595–353 B.C. 294–229 B.C. 219–213 B.C. 1290–1404 1044–1103 1118–1253 1290–1404 1415–1516 1597–1618 1278–1329 1341–1396 1439–1525 1559–1632 1450–1525 1558–1632 1455–1636 1284–1399 1415–1514 1601–1617 1447–1530 1540–1635 Probability (%) 95.4 95.4 95.4 91.5 3.9 .7 1.2 90.5 1.6 1.3 95.4 82.7 10.4 2.3 9.5 3.0 .3 76.0 6.2 .3 95.4 21.5 73.9 95.4 89.2 6.2 44.9 50.5 64.2 31.2 56.7 38.7 95.4 95.4 90.7 4.7 50.0 45.4 reflect naturally occurring, rather than culturally produced, charcoal. Alternatively, they may represent a very limited, and now deeply buried or eroded, Classic-period occupation of the site. The dates associated with the three ceramic phases do pattern chronologically, with a limited degree of overlap near the phase boundaries. Issues of overlap between phases are complicated by fluctuations in the calibration curve during the Postclassic period, which produce multiple intercepts or plateaus of approximately even probability for many of the dates. When the dates are 16 LATIN AMERICAN ANTIQUITY [Vol. 26, No. 1, 2015 Table 3. Correlations between Ceramic Phases and Radiocarbon Dates. Radiocarbon Phase Ceramic Phase Pre-Dongu Dongu Ninupi Yata D/N/Y Unknown 0 2 2 2 10 1 1 1 3 1 4 1 2 12 2 1 6 12 radiocarbon dates predate their ceramic associations (likely “old wood”); and two post-date their ceramic associations (likely contaminated contexts; see Table 3). In order to clarify the relationship between the ceramic seriation groups and absolute dates, we next turned to Bayesian methods. Bayesian Modeling Figure 6. Radiocarbon calibration curves arranged by phase. Stars mark dates that do not match the ceramic phase assignment of the associated stratigraphic layer. ordered chronologically by their single highestprobability intercept, they show a clear association with the groupings created by ceramic seriation. Of the 54 total radiocarbon dates, 39 are from contexts that could be securely placed into a ceramic phase. Of these 39 dates, 34 cases have matching ceramic and radiocarbon phases; three Bayesian methods have been rapidly gaining popularity in archaeological chronology building across a wide range of regions and time periods (Beramendi-Orosco et al. 2009; Higham and Higham 2009; Overholtzer 2014; Zeidler et al. 1998). The methods are uniquely well suited to archaeological dating, which usually includes a mix of absolutely known factors (such as historical dates), relative contextual information, and probability curves (such as radiocarbon or obsidian hydration calibrations). Bayesian methods allow for the incorporation of all of these types of data into a single model under controllable parameters. The model combines these multiple factors into a single probability curve. Curves may be generated for either the length of a phase or its beginning and end dates. In the case of Calixtlahuaca, they allow for the incorporation of ceramic seriation data, radiocarbon dates, and ethnohistoric information about the abandonment of the site into the prediction of the absolute beginning and end dates for each phase. The first two types of data have been discussed above. The third, ethnohistoric documents, provides two potentially relevant dates—the Aztec conquest of Calixtlahuaca in A.D.1476 and the abandonment of the site by Spanish decree in A.D. 1530. The source of the latter date is discussed in more detail in the next section of the paper. All Bayesian Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA analyses were performed using BCal, a publicly available Bayesian program for archaeological chronology (Buck et al. 1999). We ran four sets of Bayesian analyses to predict the beginning and end dates for each phase, referred to as Models 1–4. All four runs included three ceramic-based groups of radiocarbon dates, using only those dates that corresponded to ceramic assignments for their context (described in the previous section). This resulted in 34 initial dates, some of which were excluded from later runs due to their probability of being outliers. The date of A.D. 1476 for Aztec conquest was not included in the models after preliminary runs demonstrated that it was highly unlikely to correspond to the transition between the two Late Postclassic phases. This was not unexpected, because in other regions in central Mexico there is often significant economic and cultural interaction with the Basin of Mexico prior to actual military-political conquest (Hare and Smith 1996). The 1530 historic date for the abandonment of the site was consistent with the preliminary dates independently generated by the radiocarbon dates, so we decided to include this date in the models. It is modeled as a minimum end date for occupation at the site, on the premise that there must have been some population remaining at the site until at least 1530 to be moved by royal order. Models 1 and 2 used all 34 radiocarbon dates and assumed that the phases were independent and abutting, respectively. Models 3 and 4 removed the earliest two radiocarbon dates as outliers and, again, assumed that the phases were independent (model 3) and abutting (model 4). When the phases are independent, as in models 1 and 3, the model treats the end date of one phase as independent of the beginning of the next. This produces the six curves seen in the graphs for these models, showing the independently generated beginning and end of each of the three phases. This allowed us to check for overlaps or gaps in the time periods represented by the phases. When the phases are assumed to abut, as in models 2 and 4, the end of one phase must correspond with the beginning of the subsequent phase, and the radiocarbon dates from both periods are factored into determining the transition date. This produces a set of only four curves— the beginning of the Dongu phase, the 17 Dongu/Ninupi transition, the Ninupi/Yata transition, and the end of the Yata phase. Each analysis was run a minimum of three times to check for the effects of random noise resulting from resampling. None of the analyses produced variation of more than a few years between runs. The dates presented here are from the median run for each case, with the single highest probability years noted at the top of the figure (Figure 7). Outlier dates are a potential problem in Bayesian models. In order to test the effect of individual dates on the phase ranges, we excluded one or two dates from the ends of each phase when the distribution of dates suggested that this would be appropriate. The only case in which this produced a significant shift in phase beginning or end dates is the beginning of the Dongu phase. The exclusion of one date, R-44, results in shift of about 110 years in the start date for the occupation of the site, as can be seen by comparing runs 1 and 3 or 2 and 4 (Figure 7). Comparisons of the paired independent and abutting runs show consistent results. The highest probability beginning and end dates for the sequence as a whole remain constant. In the independent models, the maximum probability intercepts for the dates for the end of the Dongu phase and the beginning of the Ninupi phase have an approximately 50-year gap between them. Despite this, the two independent probabilities do show a large area of overlap. In the abutting model, the phase transition falls at the recent end of the date spectrum in the independent results. The independent transition between the Ninupi and Yata phases shows an opposite pattern, with an approximately 30-year zone of overlap between the end of the Ninupi phase and the beginning of the Yata phase. The abutting versions of the model place the Ninupi/Yata transition evenly between the relevant independent variables. The consistency between the independent and abutting transitions supports the validity of the date groupings used in each phase. We concluded that model 4 is the most reasonable. We chose abutting, rather than independent, phase dates for two reasons. First, there is no stratigraphic or contextual evidence for a discontinuity in site occupation between the Dongu and Ninupi phases. Second, the 30 years of overlap between the Ninupi and Yata phases are within 18 LATIN AMERICAN ANTIQUITY [Vol. 26, No. 1, 2015 Figure 7. Bayesian phase length models 1–4. the range of “fuzziness” usually accepted in phase transitions, and a single transitional date is easier to use for analytical purposes. We chose to exclude the earliest radiocarbon date from the Dongu phase due to the high probability that it was an outlier. Relationship to the Historical Chronology The central Mexican native historical chronicles contain very little information about events and processes in the Toluca Valley prior to the Spanish conquest. Tomaszewsi and Smith (2011) review mentions of Toluca Valley places in the standard Basin of Mexico sources and create a six-period chronology from the thirteenth century through the Spanish colonial period. We have followed Smith’s (1987) methodological precept of keeping the archaeological and historical chronologies separate until both have been analyzed on their own. With the partial exception of the date for the final abandonment of Calixtlahuaca (A.D. 1530), the historical chronology has played no part in the derivation of our archaeological chronology. In this section, we compare the outlines of the two chronologies (Figure 8); the full implications of correlating the two records cannot be drawn until more analyses of the excavation Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 19 Figure 8. Comparative regional chronologies. results have been carried out. Tomaszewski and Smith (2011) call the period before A.D. 1330 “Legendary” because of the difficulty of ascertaining historical accuracy from the oral traditions of native history more than two centuries prior to the Spanish conquest (Smith 2007). The archaeological remains suggest that Calixtlahuaca was founded during this period. One of the major historical processes of the thirteenth century was the arrival in central Mexico of Nahuatl speakers from Aztlan (Smith 1984). The “Matlatzinca” (which means, in this context, the “people of Matlatzinco” or “people of the Toluca Valley”) are listed as one of the arriving groups in eight of the twelve accounts of the Aztlan migrating groups. The presence of the Matlatzinca in the Aztlan lists can be interpreted two ways. First, perhaps Nahuatl-speaking groups moved into the Toluca Valley at that time, in which case the Valley was heavily multilingual for the entire existence of Calixtlahuaca. Our archaeological date for the founding of the city—A.D. 1130—fits well with the founding of many other Nahuatl city-states in central Mexico (Smith 2008). The second interpretation of the presence of Matlatzinca in the lists of Aztlan migrants is that this group was inserted into native historical accounts of the mi- 20 LATIN AMERICAN ANTIQUITY grations in order to support Mexica claims of territory and dominance in the Toluca Valley. This was the view of José García Payón (1941:19), who identified this practice as an example of the biased nature of Mexica political history. Postclassic material culture in the Toluca Valley shows more similarities with preexisting local Late Classic and Epiclassic traditions than it does with most contemporaneous Nahuatl-speaking areas. If correct, this interpretation suggests that Nahuatl speakers may not have moved into the Toluca Valley in appreciable numbers until after the Mexica conquest in A.D. 1476. The Tepanec empire (Santamarina 2006) straddles the Dongu and Ninupi archaeological phases. The historical record contains little of note during Period B, the early Tepanec empire. During the reign of the Tepanec emperor Tezozomoc (Period C, A.D. 1370–1428), several documents suggest Tepanec conquests in the Toluca Valley, as well as Tepanec ownership of some lands (Carrasco 1984; Santamarina 2006). Unfortunately, these records say little about the city of Calixtlahuaca or Matlatzinco. Period C lines up well with the Ninupi phase. Period D, corresponding to the early decades of the Mexica empire, begins with the formation of the Triple Alliance in A.D. 1428. The native historical chronicles have little to say about the Toluca Valley at this time. The next period (Period E) begins with the conquest of Calixtlahuaca and the Toluca Valley by the Mexica emperor Axayacatl in 1476. This event loomed large in the historical memory of the early colonial period, and a number of lawsuits over colonial land and labor turned on the question of political structure prior to and after Axayacatl’s conquest (García 1999; García and Jarquín 2006; Hernández 2011). These documents provide information on Calixtlahuaca’s role as capital of the Toluca Valley. They confirm that Calixtlahuaca is the same place as the city called Matlatzinco in many sources, and they provide the names of kings and dynasties, as well as other information about political dynamics at the time. After A.D. 1476, Axayacatl divided the Valley among his close relatives and political allies, and many immigrants then moved in from the Basin of Mexico. Period E is entirely contained within the Yata archaeological phase, which also includes several decades of Period D, [Vol. 26, No. 1, 2015 prior to Mexica conquest. Period E ends with the Spanish conquest of A.D. 1521. Our initial reconstructions of the Spanish colonial period suggested—from the archaeological remains alone—that the Yata phase probably continued for a decade or two after A.D. 1521. Some ceramic figurines from the uppermost levels of Yata deposits show clear European themes, yet the standard archaeological markers of early colonial central Mexican occupations at urban sites—iron artifacts, glazed pottery, bones from horse and cow—are lacking at the site. This is not a definitive argument for an absence of sixteenth century occupation, since Spanish traits did not show up at some rural Aztec sites in the Teotihuacan Valley until a century after the Spanish conquest (Charlton et al. 2005). But these new features were present at Spanish urban centers such as Yautepec and Mexico City, and Calixtlahuaca was part of the Spanish town of Toluca, included in the land grant to the conqueror Cortés (the “Marquesado del Valle”). Thus we expected to find the new material culture if there had been significant early colonial deposits at Calixtlahuaca. In addition, the Bayesian models all showed that the probability of an end date after A.D. 1530 drops very quickly. In 2013, prior to the Bayesian modeling described above, we learned of documentary evidence that the occupation of Calixtlahuaca ended around A.D. 1530 or shortly thereafter. Dr. René García Castro provided Smith with a microfilm copy of a colonial lawsuit from A.D. 1598, and Dr. Raymundo Martínez kindly arranged for the paleography of key passages. One such passage reads: The Indians who were in the lands of the capital city Calixtlahuaca later populated and built the church and monastery of San Francisco in the city of Toluca. After the Marquis had taken possession, the friars of that monastery forced the Indians who were in Calixtlahuaca to move to the city of Toluca [Marquesado 1598:f482v; translation by Smith]. The Marquis was Hernan Cortés, who received the title “Marquis del Valle de Oaxaca” in Spain in A.D. 1529 and arrived to take possession in Toluca (one of the cities included in the Marquesado, the territory associated with the title) be- Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA 21 tween A.D. 1530 and 1532. Armed with this historical date that confirmed our initial archaeological supposition, we used A.D. 1530 as the minimum end date for the Yata phase in the Bayesian modeling. Discussion The chronology described in this paper has significance for continuing research at Calixtlahuaca, for central Mexican chronologies, and for the general issue of chronological refinement in Mesoamerican archaeology. Local Significance Our new ceramic phases and their placement in time will permit a variety of inferences about the development and demise of the city and polity of Calixtlahuaca. Many of these analyses are still ongoing, and it would be premature to discuss them here. One finding that does stand out, however, is the relatively high level of apparent residential mobility within the site. As shown in Figure 9, many houses and terraces were abandoned at the end of each phase, and others were newly occupied. Although we cannot trace the movements of individual households, the most likely explanation is that households moved periodically within the urban center. This pattern presents a strong contrast to houses at contemporaneous sites in Morelos, where Smith found a very high level of continuity in house occupation across phases (Smith 1992; Smith et al. 1999). The significance of the low level of residential continuity at Calixtlahuaca is still unclear. Broadly, material culture at Calixtlahuaca became increasingly similar to that of the Basin of Mexico over time. Imported and local variants of Aztec-style ceramics increased. Ceramics indicators related to food preparation practices, such as comales and bowl-to-jar ratios, shifted toward Basin frequencies. The frequency of green obsidian increased. Nonetheless, for all of these markers, the basic domestic assemblages are still far more similar to earlier phases at the site than they are to Aztec-period domestic assemblages in the Basin of Mexico. Regional Implications Our new chronology provides for the first time a Figure 9. Continuity of occupation by excavation unit. rigorous chronology for the Middle and Late Postclassic time periods in the Toluca Valley. This advance not only improves archaeological knowledge of the Toluca area, but also helps resolve other chronological issues in central and westcentral Mexico. One of the major contributions focuses on the well-known Aztec black-on-orange typology. The dating and degree of chronological overlap among Aztec orangeware types is a long- 22 LATIN AMERICAN ANTIQUITY standing topic of investigation within central Mexican archaeology (Charlton 1996; Hodge et al. 1993; Overholtzer 2014; Parsons 1966; Vaillant 1938). Because the presence or absence of these types is often used to mark chronological change, shifts in the dating of the types have larger implications for the dating of central Mexican sites. With ongoing investigation, it has become clear that there is a temporal overlap between many of the types in the sequence, as well as some geographic variation in the periods of use. Our results provide new chronometric dates for the Aztec III and Aztec III/IV black-on-orange ceramic types and, thus, additional support for their chronological usefulness in central Mexico. The appearance of Aztec III in the Ninupi phase is slightly later than in the Basin of Mexico and in Morelos, as might be expected from a provincial region that did not participate in the development of the type. The continued use of Aztec III during the Yata Phase is consistent with sites in the Basin of Mexico. The widespread use of Aztec III/IV during the Yata phase pushes back the likely dates for the development of the type a full half-century before the date of A.D. 1500 proposed by Hodge. The earlier date is consistent with excavations in Yautepec, Morelos, where the type is present in Molotla-phase (A.D. 1440–1540) deposits (Hare and Smith 1996). This may have implications for the relative dating of material from the Basin of Mexico, where Aztec III/IV has been identified in survey (Parsons 2008) and in reconstructed historical excavations (Garraty 2013) that lack associated absolute dates. The Aztec orangeware at Calixtlahuaca firmly supports the hypothesis that the cluster of motifs characterized as Aztec III/IV is a chronologically distinctive and sensitive variant within the central Mexican Aztec orangewares. In addition to the refinement of the Aztec black-on-orange type sequence, this chronology also offers the first absolute dates for most Postclassic Toluca Valley ceramic types. While additional work will be necessary to determine regional (e.g., Sugiura 2005) or urban/rural variation in the periods of use of particular types, the general patterns presented here will allow more accurate dating of surveyed sites in the [Vol. 26, No. 1, 2015 Toluca Valley and other areas where the types appear as tradewares. One particular case where this chronology may be of interest is in the dating of the Cumbres phase, a group of site unit intrusions at the Ucareo obsidian source (Hernández and Healan 2008). Methodological Implications Our methods of chronology building are not innovative. Seriation, stratigraphic analysis, and radiocarbon dating are all time-honored methods, and Bayesian analysis of calibrated radiocarbon dates is fast becoming a standard archaeological approach. Yet for Postclassic sites in northern Mesoamerica such methods are rarely honored in practice. Lisa Overholtzer’s (2014) radiocarbon-based sequence for Xaltocan is the first new and strongly supported chronological advance for the Aztec-period Basin of Mexico in many decades. We need many more cases of targeted chronology building that incorporate radiocarbon dates, seriation, and other methods. Our results demonstrate the importance of using multiple lines of evidence to create a rigorous excavation-based chronological sequence. We attribute our success to several analytical decisions, including a focus on ceramic types rather than attributes and the use of a wide range of types in ceramic seriation, rather than one or two marker types. Furthermore, our research shows the utility of Bayesian methods in the development of refined archaeological chronologies and the potential to apply such methods outside of deeply stratified sites. Acknowledgments. The research presented in this paper was supported by NSF grants awarded to Smith for fieldwork (Urbanization and Empire at the Aztec-Period site of Calixtlahuaca, Award #0618462) and analysis (The Urban Economy of Aztec-Period Calixtlahuaca, Award #0924655), and an NSF Dissertation Improvement Grant awarded to Huster (Effects of Aztec Conquest on Provincial Commoner Households at Calixtlahuaca, Mexico, Award #1205738). Institutional support was provided by Arizona State University and El Colegio Mexiquense, where our field lab is located. We thank Arqlga. Teresa García García, Director of the Centro INAH Estado de México, for her help during the fieldwork. Juliana Novic, Aleksander Borejsza, and the rest of the archaeologists and local workers in the field crew made the project possible. Our crew of able lab assistants from the village of Calixtlahuaca completed much of the basic ceramic classification and data entry. Justin Mortensen performed the Bayesian analyses and created the resulting figures. We thank René Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA García Castro and Raymundo Martínez for help with documents and paleography. Lisa Overholtzer kindly shared an unpublished paper on the Xaltocan chronology. The comments of two anonymous reviewers and Geoffrey Braswell helped us to improve the clarity of our argument. Data Availability Statement. The radiocarbon dates analyzed in this paper are officially reported here. The ceramic type list used by the Calixtlahuaca Archaeological Project, ceramic type counts for the CSS, and discriminant analysis probabilities for each lot are available as supplemental materials accompanying this article. Spreadsheet versions of these files will be provided by the authors upon request. The complete ceramic database of the project will be uploaded to tDAR in 2015 as part of the data management plan. The artifacts recovered by the project are curated at the Colegio Mexiquense in Toluca, Estado de México, México. Supplemental Materials. Supplemental materials are linked to the online version of this paper, which is accessible via the SAA member login at www.saa.org/members-login. Supplemental Table 1. A complete list of type codes and names used by the Calixtlahuaca Archaeological Project, arranged by vessel type. Supplemental Table 2. Ceramic type counts by lot for all Core Seriation Sample (CSS lots). The table includes only those types with more than 10 sherds in the CSS. In cases where multiple types were combined for the discriminant analysis in the article, they are presented here as combined counts. A spreadsheet version of this table will be presented upon request. Supplemental Table 3. Discriminant analysis distances and group assignment probabilities for all excavated lots with more than 50 sherds. References Cited Basalenque, Fray Diego 1975 [1642] Vocabulario de la lengua castellana vuelto a la matlaltzinga. Biblioteca Enciclopédica del Estado de México Vol. 34. Toluca, México. Beramendi-Orosco, Laura E., Galia Gonzalez-Hernandez, Jaime Urrutia-Fucugauchi, Linda R. Manzanilla, Ana M. Soler-Arechalde, Avto Goguitchaishvili, and Nick Jarboe 2009 High-Resolution Chronology for the Mesoamerican Urban Center of Teotihuacan Derived from Bayesian Statistics of Radiocarbon and Archaeological Data. Quaternary Research 71:99–107. Bronk Ramsey, Christopher 2009 Bayesian Analysis of Radiocarbon Dates. Radiocarbon 51(1):337–360. Buck, Caitlin E., J. Andrés Christen, and Gary N. James 1999 BCal: An On-line Bayesian Radiocarbon Calibration Tool. Internet Archaeology 7. Carrasco, Pedro 1984 The Extent of the Tepanec Empire. In The Native Sources and the History of the Valley of Mexico, edited by J. de Durand-Forest, pp. 73–93. British Archaeological Reports, International Series, Vol. S204. Archaeopress, Oxford. Castillo Tejero, Noemí 1991 La cerámica “polícroma Matlatzinca” del viejo museo 23 de Toluca. In Homenaje a Julio César Olivé Negrete, pp. 297–324. Universidad Nacional Autónoma de México, Mexico City. Charlton, Thomas H. 1996 Early Colonial Period Ceramics: Decorated Red Ware and Orange Ware Types of the Rural Otumba Aztec Ceramic Complex. In Arqueología Mesoamericana: Homenaje a William T. Sanders, Vol. 1, edited by A. G. Mastache, J. R. Parsons, R. S. Santley, and M. C. Serra Puche, pp. 461–479. Instituto Nacional de Antropología e Historia, Mexico City. Charlton, Thomas H., Cynthia L. Otis Charlton, and Patricia Fournier 2005 The Basin of Mexico A.D. 1450–1620: Archaeological Dimensions. In The Postclassic to Spanish-Era Transition in Mesoamerica: Archaeological Perspectives, edited by S. Kepecs and R. Alexander, pp. 49–64. University of New Mexico Press, Albuquerque. Escalante Hernández, Roberto, and Marciano Hernández 1999 Matlazinca de San Francisco Oxtotilpan, Estado de México. Archivo de lenguas indígenas de México. El Colegio de México, Mexico City. García Castro, René 1999 Indios, territorio y poder en la provincia matlatzinca: la negociación del espacio político de los pueblos otomianos, siglos XV-XII. CIESAS, Instituto Nacional de Antropología e Historia and El Colegio Mexiquense, Mexico City and Toluca. García Castro, René, and María Teresa Jarquín Ortega (editors) 2006 La proeza histórica de un pueblo, San Mateo Atenco en el valle de Toluca, siglos VIII-XIX. El Colegio Mexiquense and Universidad Autónoma del Estado de México, Toluca. García Payón, José 1936 La zona arqueológica de Tecaxic-Calixtlahuaca y los matlatzincas: etnología y arqueología (primera parte). Talleres Gráficos de la Nación, Mexico City. 1941 Matlatzincas o pirindas. El Nacional, Mexico City. 1979 La zona arqueológica de Tecaxic-Calixtlahuaca y los matlatzincas: etnología y arqueología (textos de la segunda parte), edited by Wanda Tommasi de Magrelli and Leonardo Manrique Castañeda. Biblioteca Enciclopédica del Estado de México No. 30. Toluca, Mexico. Garraty, Christopher P. 2009 Attribute-Based Seriation of Postclassic and Early Colonial Sherd Collections from the Basin of Mexico. Journal of Field Archaeology 34(2):153–170. 2013 Market Development and Pottery Exchange under Aztec and Spanish Rule in Cerro Portezuelo. Ancient Mesoamerica 24(1):151–176. González Martínez, Pedro R., J. Azorín, Peter Schaaf, and Ángel Ramírez L. 1999 Assessing the Potential of Thermoluminescence Dating of Pre-Conquest Ceramics from Calixtlahuaca. Radiation Protection Dosimetry 84:483–487. Hare, Timothy S., and Michael E. Smith 1996 A New Postclassic Chronology for Yautepec, Morelos. Ancient Mesoamerica 7(2):281–297. Hernández, Christine L., and Dan M. Healan 2008 The Role of Late Pre-Contact Colonial Enclaves in the Development of the Postclassic Ucareo Valley, Michoacan, Mexico. Ancient Mesoamerica 19(2):265–282. Hernández Cruz, Luis, and Moises Victoria Torquemada 2010 Diccionario del hñähñu (otomí) del Valle del Mezquital, Estado de Hidalgo. 2nd ed. Instituto Lingüístico de Verano, Mexico City. 24 LATIN AMERICAN ANTIQUITY Hernández Rodríguez, Rosaura 2011 La tradición oral en el valle de Toluca. In Historia general ilustrada del Estado de México, tomo 2, etnohistoria, edited by R. Hernández Rodríguez and R. C. Martínez García, pp. 179–197. El Colegio Mexiquense, Toluca, México. Higham, Charles, and Thomas Higham 2009 A New Chronological Framework for Prehistoric Southeast Asia, Based on a Bayesian Model from Ban Non Wat. Antiquity 83(319):125–144. Hodge, Mary G. 1998 Archaeological Views of Aztec Culture. Journal of Archaeological Research 6:197–238. 2008 Un lugar de jade: sociedad y economía en el antiguo Chalco/Place of Jade: Society and Economy in Ancient Chalco. Instituto Nacional de Antropología e Historia and University of Pittsburgh. Mexico City and Pittsburgh. Hodge, Mary G., and Leah D. Minc 1991 Aztec-Period Ceramic Distribution and Exchange Systems. Final Report submitted to the National Science Foundation. Copy on file at the Department of Anthropology, Oregon State University. Hodge, Mary G., Hector Neff, M. James Blackman, and Leah D. Minc 1993 Black-on-Orange Ceramic Production in the Aztec Empire’s Heartland. Latin American Antiquity 4(2):130– 157. Kiemole Muro, Mildred 1975 Vocabulario mazahua-español y español-mazahua. Biblioteca Enciclopédica del Estado de México Vol. 35. Toluca, México. Lastra de Suárez, Yolanda 1989 Otomí de San Andrés Cuexcontitlán, Estado de México. El Colegio de México, Mexico City. 2008 Topónimos otomíes. Estudios de Cultura Otopame 6:381–314. Marquesado 1598 El fiscal de su magestad contra el marquez del Valle sobre los pueblos de Toluca. Documento del Archivo General de la Nación, Hospital de Jesus, leg. 277, Exp 2. Nichols, Deborah L,. and Thomas H. Charlton 1996 The Postclassic Occupation at Otumba: A Chronological Assessment. Ancient Mesoamerica 7:231–244. Overholtzer, Lisa 2014 A New Bayesian Chronology for Postclassic and Colonial Occupation at Xaltocan, Mexico. Radiocarbon 56(3):1077–1092. Parsons, Jeffrey R. 1966 The Aztec Ceramic Sequence in the Teotihuacan Valley, Mexico. 2 vols. Ph.D. dissertation, Department of Anthropology, University of Michigan, Ann Arbor. 2008 Prehispanic Settlement Patterns in the Northwestern Valley of Mexico: The Zumpango Region. Memoirs No. 45. Museum of Anthropology, University of Michigan, Ann Arbor. Piña Chán, Román 1981 Investigaciones sobre Huamango y región vecina (Memoria del Proyecto). 2 vols. Dirección de Turismo del Gobierno del Estado de México, Toluca. Reimer, P. J., M. G. L. Baillie, E. Bard, A. Bayliss, J. W. Beck, P. G. Blackwell, C. Bronk Ramsey , C. E. Buck, G. S. Burr, R. L. Edwards, M. Friedrich, P. M. Grootes, T. P. Guilderson, I. Hajdas, T. J. Heaton, A. G. Hogg, K. A. Hughen, K. F. Kaiser, B. Kromer, F. G. McCormac, S. W. Manning, R. W. Reimer, D. A. Richards, J. R. Southon, S. Talamo, C. S. M. Turney, J. van der Plicht, and C. E. Weyhenmeyer [Vol. 26, No. 1, 2015 2009 INTCAL 09 and MARINE09 Radiocarbon Age Calibration Curves, 0–50,000 Years cal BP. Radiocarbon 51(4):1111–1150. Rodríguez Villegas, Manuel 2010 Spanish-Mazahua Dictionary On-line in AULEX. AULEX, Mexico City. Electronic document, http://aulex.org/ es-maz/, accessed April 11, 2014. Santamarina, Carlos 2006 El sistema de dominación azteca: el imperio tepaneca. Fundación Universitaria Española, Madrid. Smith, Michael E. 1983 Postclassic Culture Change in Western Morelos, Mexico: The Development and Correlation of Archaeological and Ethnohistorical Chronologies. Ph.D. dissertation, Department of Anthropology, University of Illinois. 1984 The Aztlan Migrations of the Nahuatl Chronicles: Myth or History? Ethnohistory 31:153–186. 1987 The Expansion of the Aztec Empire: A Case Study in the Correlation of Diachronic Archaeological and Ethnohistorical Data. American Antiquity 52:37–54. 1992 Archaeological Research at Aztec-Period Rural Sites in Morelos, Mexico. Vol. 1, Excavations and Architecture/Investigaciones Arqueológicas en Sitios Rurales de la Época Azteca en Morelos, Tomo 1, Excavaciones y Arquitectura. University of Pittsburgh Memoirs in Latin American Archaeology 4. University of Pittsburgh, Pittsburgh. 2001 Postclassic Ceramics from the Toluca Valley in U.S. Museums: The Bauer and Blake Collections. Mexicon 23:141–146. 2007 Tula and Chichén Itzá: Are We Asking the Right Questions? In Twin Tollans: Chichén Itzá, Tula, and the Epiclassic to Early Postclassic Mesoamerican World, edited by J. Kowalski and C. Kristan-Graham, pp. 579– 618. Dumbarton Oaks, Washington, D.C. 2008 Aztec City-State Capitals. University Press of Florida, Gainesville. Smith, Michael E., Aleksander Borejsza, Angela C. Huster, Charles D. Frederick, Isabel Rodríguez López, and Cynthia Heath-Smith 2013 Aztec Period Houses and Terraces at Calixtlahuaca: The Changing Morphology of a Mesoamerican Hilltop Urban Center. Journal of Field Archaeology 38(3):225– 241. Smith, Michael E., and John F. Doershuk 1991 Late Postclassic Chronology in Western Morelos, Mexico. Latin American Antiquity 2:291–310. Smith, Michael E., Cynthia Heath-Smith, and Lisa Montiel 1999 Excavations of Aztec Urban Houses at Yautepec, Mexico. Latin American Antiquity 10(2):133–150. Smith, Michael E., and Michael Lind 2005 Xoo-Phase Ceramics from Oaxaca Found at Calixtlahuaca in Central Mexico. Ancient Mesoamerica 16:169– 177. Smith, Michael E., Juliana Novic, Angela C. Huster, and Peter C. Kroefges 2009 Reconocimiento superficial y mapeo en Calixtlahuaca en 2006. Expresión Antropológica 36:39–55. Smith, Michael E., Jennifer B. Wharton, and Melissa McCarron 2003 Las ofrendas de Calixtlahuaca. Expresión Antropológica 19:35–53. Stoner, Wesley, and Michael D. Glascock 2013 Neutron Activation Analysis of Pottery from the Toluca Valley. Archaeometry Laboratory, Missouri University Research Reactor. Report submitted to M.E. Smith and A. Huster. Copies available from M.E. Smith or A. Huster, Arizona State University. Huster & Smith] A NEW ARCHAEOLOGICAL CHRONOLOGY FOR AZTEC-PERIOD CALIXTLAHUACA Sugiura Yamamoto, Yoko 1998 El valle de Toluca después del ocaso del estado Teotihuacano: el Epiclásico y el Posclásico. In Historia general del Estado de México, vol. 1: geografía y arqueología, edited by Yoko Sugiura Yamamoto, pp. 199–259. Gobierno del Estado de México and El Colegio Mexiquense, Toluca. 2005 Reacomodo demográfico y conformación multiétnica en el valle de Toluca durante el Posclásico: una propuesta desde la arqueología. In Reacomodos demográficos del clásico al posclásico en el centro de México, edited by L. Manzanilla, pp. 175–202. Instituto de Investigaciones Antropológicas, Universidad Nacional Autónoma de México, Mexico City. Tomaszewski, Brian M., and Michael E. Smith 2011 Politics, Territory, and Historical Change in Postclassic Matlatzinco (Toluca Valley, Central Mexico). Journal of Historical Geography 37:22–39. Tommasi de Magrelli, Wanda 1978 La cerámica funeraria de Teotenango. Biblioteca Enciclopédica del Estado de México 61. Toluca, Mexico. 25 Vaillant, George C. 1938 A Correlation of Archaeological and Historical Sequences in the Valley of Mexico. American Anthropologist 40:535–573. Vargas Pacheco, Ernesto 1975 La cerámica. In Teotenango: el antiguo lugar de la muralla, Vol. 2., edited by Román Piña Chan, pp. 189– 264. Gobierno del Estado de México, Mexico City. Zeidler, James A., Caitlin E. Buck, and Clifford D. Litton 1998 Integration of Archaeological Phase Information and Radiocarbon Results from the Jama River Valley, Ecuador: A Bayesian Approach. Latin American Antiquity 9(2):160– 179. Submitted April 11th, 2014; Revised August 22nd, 2014; Accepted September 11th, 2014.