- Social Sciences
- African Studies
- American Studies
- Asian Studies
- Communication Sciences
- Ethnic Studies
- European Studies
- Gender Studies
- Physical Sciences
- Life Sciences
- Animal Communications
- Cell Biology
- Evolutionary Biology
- Food Science and Technology
- Human Anatomy
Correspondence in Tree-Ring and Thermoluminescence Dating: A Protohistoric Navajo Pilot Study
by James K. Feathers, Ronald H. Towner, Douglas D. Dykeman
Correspondence in Tree-Ring and Thermoluminescence Dating: A Protohistoric Navajo Pilot Study
James K. Feathers, Ronald H. Towner, Douglas D. Dykeman
Updated: November 5th, 2012
CORRESPONDENCE IN TREE-RING AND THERMOLUMINESCENCE
DATING: A PROTOHISTORIC NAVAJO PILOT STUDY
Douglas D. Dykeman, Ronald H. Towner, and James K. Feathers
Dating of early iVavajo resiclence and special use sites, en. A.D. 1500-1775, has been hampered by a lack of clatable nzateri- als and poor precision in radiocarbon results. Methocls described in this paper use materials ubiquitous at early Navajo sites in northwestern New Mexico and ernploy a dual strategy involving tree-ring dating of nonarchitectural wood and thermolumi- nescence assay of ceramics and burneel rock. Comparison of samples obtained from a number of sites near the Morris Site I pueblito indicates remarkable correspondence between tree-ring ancl thermolzilminescence results. These techniques are argued to have considerable reliability for relatively recent cultural manifestations such as these early Navajo sites. Thennolumines- cence in particular may be usefill in protohistoric contexts where tree-ring elating is unavailable. The thennoluminescence tech- nique has the added benefit of directly dating pottery shercls, which can be useful for cleveloping ceramic cross-dating sequences.
Los fechados arqueoldgicos de 10s lugares de residencia y sitios de uso especial de 10s primeros Navajo, entre 10s afios 1500 y 1775 en el noroeste de iVuevo Me'xico, han estado incompletos por la falta de objetos fechados y la pobre precisidn de 10s resul- taclos cle la ticnica cle racliocarbdn. Los me'todos descritos en este clocumento utilizan lnateriales encontrados por cualquierparte de 10s lugares de 10s primeros Navajo, en una doble estrategia que incluye la datacidn por 10s anillos de 10s drboles reflejados en la nzaclera cle 10s mismos y el ensayo de termoluminiscenia de cerdmica y roca quemada. La comparacidn de lnuestras obtenidas en lugares cercanos a1 pueblito Morris Site I, indican corresponclencias notables entre 10s resultados de ambas te'cnicas: la de 10s anillos cle 10s drboles y la de termoluminiscenia. Por consiguiente, estas tecnicas son argumentadas para tener considerable veraciclad de las manifestaciones culturales relativarnente recientes, corno 10s primeros lugares de 10s iVavajo. La ternzolu- miniscenia en particular puede ser itil en contextos prehistdricos doncle la clatacidn de 10s anillos de 10s drboles no es posible. La ticnica de termoluminiscenia tiene el benej7cio ariadido de fechar directamente 10s fragmentos de cerdmica, lo que puede ser litil para el desarrollo de la secuencia de datos a trave's de la cerdnzica.
ating protohistoric sites is difficult because site contexts in a single contiguous project area,
of the need to distinguish closely timed known as The Morris Site 1 Early Navajo Land Use
events in the interpretation and synthesis of Study.
the recent past. Traditional archaeometric dating
Protohistoric Navajo and the Morris
techniques, with the exception of tree-ring dating,
Site 1 Project
have not improved on the rather coarse resolution provided by diagnostic material culture. Relying on The Morris Site 1 project is centered in Dintitah, tra- the latter is not satisfactory because of the rarity or ditional homeland of the Navajo in northwestern lengthy production period of some diagnostics and New Mexico, which is one of the most intensive pro- the fact that their time frames are determined by the tohistoric Navajo occupations known in the South- same archaeometric techniques. In this paper we west (Figure 1). Navajo use of the area occurred evaluate the resolving power of thennoluminescence primarily during the Early Navajo period (ca. A.D. (TL) dating, which is just beginning to see applica- 1450-1863), which is composed of three phases: tion in North American archaeology (Feathers 1997 DinCtah phase (A.D. 1450 [or earlier]-1625), Gob- and 2000a).' The accuracy and precision of TL dates ernador phase (A.D. 1625-1775), and Cabezon on ceramics and burned stone are judged by com- phase (A.D. 1776-1863) (Hester 1962; Hogan et al. parison with tree-ring dates obtained from the same 1991; Marshall 1995). Early Navajo period, as used
Douglas D. Dykeman .Navajo Nation Archaeology Department, 609 S. Behrend Ave., Farmington, NM 87401
Ronald H. Towner .Laboratory of Tree-ring Research, University of Arizona, Tucson, AZ 85721
James K. Feathers .Thermoluminescence Laboratory, University of Washington, Seattle, WA 98195
American Antiquity, 67(1), 2002, pp. 145-164
Copyright0 2002 by the Society for American Archaeology
Figure 1. The Morris Site 1 project area in Dinetah in the Four Corners area.
here, is virtually synonymous with Navajo period (Dittert el al. 1961), Early Agricultural and Devel- opmental Herding periods (Bailey and Bailey 1986), and several historical time frames discussed by Reeve (1957, 1958, 1959).
The DinCtah phase represents Navajo entry into the Southwest during which the Navajos established relations with the Pueblos that were both coopera- tive and confrontational (Reeve 1957, 195 8). Strongly influenced by Pueblo practices, farming was added to a diverse Navajo economy that already included hunting, foraging, raiding, and trading (Brugge 1964; Dykeman 1999). Dindtah Gray pot- tery, a locally produced sand-tempered plain ware, is associated with this phase but continued to be made in the Gobernador and early Cabezon phases. Puebloan trade wares like Jemez Black-on-white and several varieties of Rio Grande glaze wares are often found on DinCtah phase sites. Navajo settlement at this time consisted of small residential units com- posed of one or more conical forked-stick hogans constructed of wood and mud. Such sites were dis- tributed along the San Juan River and extended from the La Plata River eastward, perhaps as far as the
Chama valley (Brown 1996; Goodman 1982).
The Gobernador phase opens with rather con- certed efforts by Franciscan missionaries to convert Navajo populations in DinCtah (Hodge et al. 1945). Navajo cultural change is largely the result of inter- action with both Spaniards and Pueblos of Northern New Spain. Sheepherding was added to the already diverse subsistence practices, and the introduction of horses had rather dramatic effects on all parts of the economy (John 1975). These changes are evident in Navajo material culture assemblages that contain a variety of European goods including, but not limited to, metal axes and knives, tack, crockery, and trade beads (Carlson 1965; Hester 1962). Local produc- tion of Gobernador Polychrome begins early and ceases late in the Gobernador phase, making it an
excellent time marker for archaeologists. Puebloan trade-ware pottery associated with the Gobernador phase includes Hopi yellow ware, Jemez Black-on- white, Rio Grande glaze ware, and matte paint vari- eties from Zuni, Acoma, and the Rio Grande area. Pueblitos, multiple-room stone masonry structures that function for storage, defense, and residential purposes, were added to an increasingly concen- trated population of hogans. Gobernador phase set- tlements are concentrated south of the San Juan River in the vicinity of Gobernador and Largo canyons. Some population dispersal is evident in the latter half of the eighteenth century as indicated by outly- ing Navajo settlements found as far west as Ganado, Arizona and as far south as Cebolleta in the Mount Taylor region of New Mexico (Gilpin 1996; Good- man 1982; Hester 1962; York and Winter 1988).
The Cabezon phase features the final breakup of DinCtah Navajo communities and a shift in both set- tlement pattern and subsistence practices. Navajo communities were loosely reorganized in four or five separate areas stretching from the Mount Taylor region westward to the vicinity of Ganado and north- ward to the Chuska Mountains (Goodman 1982; York and Winter 1988). This realignment may have been the result of a strong economic shift toward sheepherding and increased hostilities between Navajos and Spaniards (Dykeman 1999; Haskell
1987; McNitt 1972). Mexico won independence
from Spain in 1821, and administrative control of
Northern New Spain remained in Mexican hands
until military defeat by the United States in 1846.
Navajos retained their status as an independent nation
during both the Spanish and Mexican administra-
tions. However, tiring of Navajo and Apache raid-
ing, in 1863 the U.S. Government moved them to
the Bosque Redondo reservation in eastern New
Mexico. The loss of Navajo independence and sov-
ereignty in 1863 marked the end of Cabezon phase
and the Early Navajo period.
The Morris Site 1Early Navajo Land Use Study is part of a massive Bureau of Land Management (BLM) undertaking known as the Fruitland Data Recovery Program, the purpose of which is the mit- igation of effects associated with the construction of an extensive gas pipeline gathering system in north- western New Me~ico.~
The BLM requested alterna- tive mitigation plans, like the Morris Site 1 project, which more effectively addressed archaeological issues, located in the vicinity of pipeline development projects. The community-oriented research design prepared for the land-use study required good tem- poral control of the site population to assure con- temporaneity of the targeted settlement pattern (Dykeman and Wharton 1994). Consequently, one of the research questions was oriented toward obtain- ing absolute dates and evaluating the utility of the results.
The Morris Site 1 project entailed extensive archaeological survey and excavation in a 567-ha project area that surrounds a small Gobernador- phase pueblito excavated by Earl Morris in 1915 (Carlson 1965). Morris designated the pueblito "Site 1" in his notes, and although assigned site number LA 83529, archaeologists continue to refer to it as Morris Site 1. In total, 121 Navajo sites, including 2 pueblitos (Morris Site 1 and Romine Canyon Ruin), 50 residential sites, and 69 special use sites, are recorded in the project area. Two sites were com- pletely excavated and sample excavation units were placed in the middens at 11 sites. The sample exca- vation units were designed expressly for the purpose of obtaining control TL samples from subsurface contexts. Tree-ring and TL samples were recovered from surface contexts at sites throughout the pro- ject area.
Pueblito sites often exhibit good preservation of wood (Towner 1997; Towner and Johnson 1998), therefore it was reasonable to expect the recovery of suitable tree-ring samples from this site type. Unfor- tunately, pueblitos represent a small proportion of the total site population, and although additional tree- ring dates could be obtained from hogans, wood preservation necessary for tree-ring dating at these open sites could not be assured. Tree-ring dating pro- vides reliable temporal control desired for commu- nity studies, but the difficulties of obtaining well-preserved samples limit the utility of the tech- nique for extensive inventory projects. Thermolu- minescence dating of surface collections has the potential for improving the dating of sites found in
inventory projects (Dunnell and Feathers 1995), but its accuracy has been questioned (Brown et al. 1992). Consequently, a section of the project data recovery plan was oriented toward establishing confidence in the TL technique by evaluating its correspondence with tree-ring results.
Radiocarbon, ceramic cross dating, and tree-ring dat- ing are valuable techniques because they date mate- rials commonly associated with archaeological sites in the Southwest. The resultant dates, however, must be relevant to the archaeological problem at hand. Dean (1978) distinguished the "target event," which is the event of interest to the archaeologist, and the "dated event," which is the event addressed by the particular dating technique. Where these events do not coincide, bridging events or arguments are needed to link the two. Each technique has strengths and weaknesses when applied to archaeological problems.
The radiocarbon technique often produces dates not relevant to the past use of the site. Radiocarbon age consistently overestimated target events by 200- 300 years in Lomolai phase (Basketmaker 11) sites on Black Mesa (Smiley and Ahlstrom 1997). Fet- terman (1996) demonstrated similar error ranges in a comparative study of radiocarbon dates obtained from wood charcoal and carbonized annual plants recovered from early Navajo sites. He found that the vast majority of wood charcoal samples predate A.D. 1500, which is at least 200 years earlier than the expected ages indicated by tree-ring results. Other studies have shown that radiocarbon dates consis- tently overestimated the age of DinCtah-phase sites by 150 years (Brown and Hancock 1992; Dykeman 2000). Simply adding 150-200 years to all dates is unsatisfactory in the absence of a rational bridging argument (Dean 1978; Smiley and Ahlstrom 1997).
Fetterman (1996) suggests that some of the age overestimation is attributable to old wood use and the cross-section effect (sometimes referred to as "built-in age" [see Hogan 19891). Such problems are associated with radiocarbon dates derived from wood charcoal. Consequently, a potential solution is to date annual plant remains, like corn. In Fetterman's case study, dates derived from annuals appear to be more accurate, but produced calibrated ranges of more than 400 years (Fetterman 1996). This range spans much of the late prehistoric period, all of the proto- historic period, and much of the historic period. Fet- terman's case is not unique, because similar ranges have been reported from other excavated Navajo sites in the area (Ayers and Reed 1993; Dykeman and Wharton 2000; Kotyk and Cater 1998; Wharton et
a1.1996). Thus, in the protohistoric period, the poor resolution often returned by radiocarbon-dated annu- als is too coarse for most archaeological site-inter- pretation purposes.
Ceramic cross dating is a time-honored site-dat- ing method in the Southwest because change in ceramic style and technology has temporal sensitiv- ity (Blinman 2000; Breternitz 1966; Colton 1953). The technique requires a tree-ring-dated sequence of discrete and recognizable ceramic types that can be used to approximate the age of otherwise undated sites. The similarity between the ceramic assemblage of a site and ceramic types keyed to a tree-ring-dated sequence provides the cross-reference for determin- ing site age. Primary causes for error in ceramic cross dating are unsupported dates for ceramic assem- blages and a lack of temporal sensitivity due to lit- tle or no change in style or technology. Also, dating resolution varies directly with the production period of individual ceramic types and may be affected by pottery use-life that substantially postdates the time of production.
Ceramic cross dating of early Navajo assemblages tends to suffer from both these dating problems. First, although ceramic variety is apparent at some sites, only two ceramic types are commonly found at early Navajo sites: DinCtah Gray and Gobernador Poly- chrome. DinCtah Gray is the most common ceramic type, but its technology remained largely unchanged from about A.D. 1500-1800 (Brugge 1981; Hill
1995). Consequently, cross dating by the occurrence
of DinCtah Gray has resolving power only slightly
better than calibrated radiocarbon dates.
The production of Gobernador Polychrome
ranges from A.D. 1640-1800 (Brugge 1981; Lan-
genfeld 1999; Marshall 1995; Reed and Reed 1996),
and thus has potentially good resolving power for
ceramic cross dating. Unfortunately, it is rarely abun-
dant and sometimes absent from Navajo sites known
to date in the A.D. 1700s. Therefore, although the
presence of Gobernador Polychrome is a good indi-
cator of a seventeenth- or eighteenth-century site, its
absence in site contexts is not necessarily indicative
of an earlier (or later) occupation. Ceramic cross dat-
ing with sufficient precision to distinguish Gober-
nador phase can be successful only when Gobernador
Polychrome is present at early Navajo sites.
Relief from problems of poor resolution and accu-
racy can be found in tree-ring dating. The ability of
the technique to produce relevant dates is dependent
upon recovery of well-preserved wood samples.
Wood is perishable, but can be found intact, or nearly
so, in the roofs of Navajo pueblito structures (Towner
1997). This does not fully resolve the dating prob-
lem as these stone structures represent less than 2
percent of Navajo sites in the Morris Site 1 project;
this value may accurately reflect the density of
pueblitos in the whole of DinCtah. Most residential
sites contain forked-stick hogans that are burned or
dismantled, or in a collapsed and eroded state that
Morris Site 1 Project Tree-ring and TL Dates
@ Tree-ring Date V TL Date
@ Tree-ring & TL Dates \ V I
\ Project Area Boundary \I
Figure 2. Map showing the spatial distribution of dated sites in the Morris Site 1 project area.
make tree-ring dating ineffective.There is little direct occupation. These factors in combination with the evidence indicating the temporal relationship of large project size and high early Navajo site density pueblitos to the vast majority of residential and non-indicated that we could be reasonably assured of residential sites, therefore, there is little basis for obtaining suitabletree-ring andTL samplesin surface extending the pueblito dates to the greater settlement and near-surface contexts (Figure 2). system. Establishing this relationship is critical for The primary goal of the tree-ring and TL comboth the land-use study and broader issues of early parison is to evaluate the accuracy and precision of Navajo chronology. TL dating results in terms of the resolution possible
Thermoluminescencedating showsgood potential for protohistoric sites. The tree-ring results are used in this regard because it only requires samples of as the standard for accuracy for identifying the tar-burned cultural materials from surface or subsurface get event. Consequently, the accuracy of Morris Site contexts (Dunnelland Feathers 1995).This technique 1 project tree-ring dates is critically evaluated to is widely used in Europe but has received less atten-assure that only valid dates are applied to the comtion in the United States, in part because of cost and parison. Valid, in this sense, means dating results questions about accuracy and precision (Feathers that represent individual tree-harvesting events or 2000a). These can be tested by evaluatingthe level of date clusters that indicate construction of site feacorrespondence between TL results and secure tree-tures, or both. TL results are not expected to agree ring dates. The character of the Monis Site 1project completely with the tree-ring dates because the two area and the nature of the early Navajo occupation are techniques date different events: the death of a tree not only suitable for tree-ring and TL techniques, but in the case of tree-ring dating and the heating of an may provide nearly ideal conditions to perform this object in the case of TL. However, early Navajo sites test. The use of both dating techniques is facilitated represent relatively short occupations and the probby (1)the customary use of both wood and pottery by ability is high that wood use and firing of objects the early Navajo inhabitants, (2) the good preserva-(either ceramic or stone) are reasonably close in time. tion of exposed perishable materials combined with We assume for now that wood and burned objects the lack of significant postabandonment disturbance found together at the same site represent contempoof sites, and (3)the relativelyrecent time frame of the rary events in an archaeological time scale,but revisit
this assumption later on. First we evaluate the tree- ring and TL results and then consider their corre- spondence.
Two types of tree-ring samples, architectural and arboreal wood, were collected to date the early Navajo sites in the Morris Site 1 project area. Archi- tectural wood was collected from buildings and other archaeologicalfeatures. Arboreal wood was collected from trees that have been modified by cultural activ- ities (Towner et al. 1998). All samples were obtained from surface contexts on or adjacent to recorded archaeological sites. Samples extracted from archi- tectural wood, the traditional method for den- drochronological samples (Stokes and Smiley 1968), were somewhat scarce but were obtained from the remains of sweat lodges, collapsed forked-stick hogans, or pueblitos.
Sampling culturally modified trees is an alterna- tive method for obtaining tree-ring samples that has been developed for early Navajo sites in the Dink- tah area (Towner et al. 1998). This method entails extraction of samples from axe-cut stumps, limbs, and tree boles, both living and dead, in the vicinity of early Navajo sites (Towner and Johnson 1998; Towner et al. 1998).
Both the tools and methods used by early Nava- jos to harvest wood allowed field archaeologists to accurately identify trees modified by such cultural activity. Iron axes of European manufacture first became available to Navajos between A.D.
1600-1625, obtained through trade or other means from the Spanish colony of New Spain (Carlson 1965; Dykeman 1999; Hester 1962). The composi- tion of the iron in these axes was rather brittle and the axe head was both smaller and lighter than Amer- ican-made axes introduced in the middle part of the nineteenth century-nearly 100 years after the pur- ported abandonment of Dinktah by Navajos. The less-durable European axes could not withstand the force of a long chopping motion that develops great head speed. A less-forceful chopping motion char- acterized by short swings probably best describes the technique required for using such iron axes. Conse- quently, this technique produces a distinctive "stepped" chip-scar pattern that is easily distin- guished from the large chip scars and V-shaped cut produced by the later American axes. Finally, the Morris Site 1 project area is rather isolated and shows
little evidence of modern wood collecting. Thus, most culturally modified trees can be attributed to the Navajo occupation of the area (Towner 1999).
Another important circumstance of early Navajo tree harvesting was that tree trunks were often cut at about 1 m above the ground, a convenient height for a standing person to wield an axe. This was fortu- itous for archaeological tree-ring sampling, because the removal of the upper part of the bole did not always kill the tree, particularly in the case of the juniper (Juniperus spp.) trees that are important in the Morris Site 1 area ecosystem. The lower portion of juniper stumps occasionally survived this har- vesting event and continued to grow branches and secondary trunks. In enough instances, this new growth developed around the dead portion of the meristem and protected the tree rings from weath- ering. Thus protected, stumps often exhibit the least amount of ring loss due to erosion and sometimes provide accurate death dates for the main bole. Such death dates correspond to harvesting events, which
are related by spatial association to the nearby site. In fact, most of the axe-cut stumps and limbs sam- pled for this project occurred within site boundaries established by feature and artifact distributions, and likely reflect wood harvesting for hogan construc- tion or fuel use.
Tree-ring samples obtained from stumps and limbs are less likely to yield anomalous dates due to the old-wood effect, because the death of the tree or meristem is directly related to the harvesting event. The dates may be anomalous because of exterior ring loss due to natural or cultural factors, but such a determination can only be made in an archaeolog- ical and dendrochronological context. If standing deadwood was harvested in this fashion, the resul- tant date would reflect the old-wood effect. Dead- wood tends to be very hard, however; the difficulty of felling a dead tree, and possibility of axe break-
age suggest that such procurement was rare.
Despite the care taken to obtain datable samples, the majority of the tree-ring dates from the Morris 1 project area are neither death nor near-death dates of trees. Most samples suffered deterioration and ero- sion from weathering to the extent that some loss of exterior tree rings was indicated. This loss always yielded an outside ring on the sample that predated the actual death date of the tree and, as a conse- quence, predated cultural use of the wood. Protocols established by the Laboratory of Tree-ring Research in the 1960s (Bannister et al. 1966) denote the con- dition of the sample and probability that the last ring on the sample is the last grown by the tree. Cutting dates (symbols B,G,L,r,c,v) indicate the year of tree death, whereas noncutting dates (vv) do not; "+" dates indicate a potentially locally absent ring near the end of the ring sequence and, if combined with a cutting date symbol, may be considered near cut- ting dates; "++" dates indicate that beyond a certain point the sample does not cross date, but is a ring count that, even if combined with a symbol of a true outside ring (B,G,L,r,c,v), cannot be interpreted as
a death date.
The Laboratory of Tree-ring Research does not routinely conduct analysis to estimate tree death dates of samples classified as noncutting dates. In some cases, however, additional analysis may indi- cate a range of possible dates. Several tree species including Douglas-fir, ponderosa pine, juniper, and oak exhibit distinctive visual differences between heartwood and sapwood rings. Heartwood forms at a certain age when cell walls become lignified and acquire a distinctive dark color. Heartwood is phys- iologically dead and serves as structural support for the tree, but is unable to transport fluids from the roots. Sapwood, which is characterized by thin- walled, open-lumined rings that are light in color, has ceased to produce new cells through cell division, but provides the mechanism for the transport of flu- ids. The outside sapwood ring is the only ring on the tree that grows; it indicates the last year of tree growth. It is this outside ring that is first subject to
deterioration upon the death of the tree. Analysis of
the character of the heartwood and sapwood, and the
boundary between the two, may provide chronolog-
ical information permitting an estimation of the num-
ber of sapwood rings lost due to deterioration.
Several attempts have been made to quantify the
number of sapwood rings typically produced by indi-
vidual tree species (Douglass 1939; Graves 1991;
Nash 1997; Plog 1980) and use that value as a con-
stant for estimating age from the heartwood-sap-
wood boundary. Such a technique has been used
successfully in Europe where oak is the dominant
tree used in dendrochronology (Bailey 1995). Using
regression analysis, Nash (1997) has provided sta-
tistical estimates of cutting dates for Douglas-fir and
ponderosa pine samples that exhibit pith dates and
Nash's (1997) methods were tested on the Mor-
ris Site 1 tree-ring collection and produced erratic results (Towner 1999). Unfortunately, arigorous sta- tistical approach was not possible with the Morris Site 1 samples because of the tree species used. PiAon pine often does not form visually distinctive heart- wood, because the cells of the interior rings rarely become lignified. Juniper presents somewhat of the opposite problem because it has the tendency to become heavily lignified, but the heartwood-sap- wood boundary is typically wavy or indistinct. More- over, the frequency of sapwood rings put on after the boundary varies tremendously between individual trees. Any statistical calculations based on the heart- wood-sapwood ratio of pifion or juniper, therefore, may have unacceptably large margins of error. The most important point to remember is that any sap- wood estimate can only add rings or years to non- cutting dates. For example, a noncutting date of
1600vv can only date after 1600 due to an estimate of sapwood ring loss.
In the absence of a viable statistical method to evaluate noncutting dates, a site-based tree-ring analysis was employed. In order to examine indi- vidual sites, each site must have yielded a large enough sample of dates to assess date clustering, cutting-date distributions, etc. (Table 1). Twenty- nine occupational components at 25 sites were tree- ring dated by this kind of analysis. In addition, six episodes of tree harvesting were documented at five stumps recorded as isolated manifestations (IM)not spatially associated with sites. Unfortunately, nine sites did not meet these criteria and could not be ana- lyzed chronologically. Sites with few dates, partic- ularly if they are noncutting dates, are much more difficult to assess in terms of their temporal occupa- tion span. The analysis of these sites is based on the number of cutting dates, clusters of cutting and non-
cutting dates, and a qualitative microscopic assess- ment of the sapwood ring count and condition of exterior rings on samples. Clustering in this context refers to individual sites with multiple tree-ring dates that collectively support one or more tree-harvest- ing or construction episodes. The construction of a stem-and-leaf plot is sometimes necessary to evalu- ate large groups of dates from a single site. Although interpretations are based on these criteria, inferences regarding sites and samples that produced only non- cutting dates should be considered tentative and somewhat speculative; they provide a date before which an event could not have occurred, but cannot
delineate a terminus post quem (translated as "date after which"). Nonetheless, most of the estimates are probably accurate within a decade, which is an acceptable tolerance for dating archaeological sites. An evaluation of confidence for tree-ring dates esti- mated in this manner is provided in Table 1. Good confidence may be assumed for interpreted dates resulting from specimens that exhibit a heartwood boundary and some preserved sapwood rings. Low confidence is indicated for specimens that lack sap- wood or exhibit other problems at the heartwood- sapwood boundary. Specimens listed as a cutting date are of the highest confidence possible in tree- ring dating.
The estimated tree-ring dates indicate two main periods of tree harvesting. The historic period is rep- resented by four dates in the late nineteenth century and early twentieth century, which corresponds well with an increase in homesteading during the Lucero phase (A.D. 1870-1960) (Dittert et al. 1961). Inter- estingly, two dates in the early twentieth century may be related to Earl Morris's work in the Gobernador area. Navajo tree harvesting in the Morris Site 1 pro- ject area is indicated by a series of dates beginning atA.D. 1629 and ending inA.D. 1750. The 121-year range indicated for the early Navajo occupation occurs entirely within the Gobernador phase that dates fr0mA.D. 1625 to about A.D. 1775 (Hogan et al. 1991; Marshall 1995; Reed and Reed 1996).
Thermoluminescence dating is based on the accu- mulation of stored energy in some crystalline mate- rials as afunction of natural radioactivity. This energy can be released by heat, resulting in the emission of light, or luminescence. The luminescence signal is proportional to the amount of stored energy, which in turn is proportional to the amount of absorbed radi- ation since the last heating event. Knowledge of the rate at which this energy was absorbed and the amount of radiation required to produce the natural signal, both measurable quantities, can lead to an esti- mate of the time since last heating (Aitken 1985). This heating must be sufficient to remove any pre- vious accumulation of energy, i.e., to reset the lumin- escence clock. This requirement is usually met by ceramics during their manufacture, if not their use, and may be met by stones from hearths.
Pottery and burned sandstone are common at early Navajo sites and provide suitable samples for TL dat-
ing. Either surface or subsurface samples can be used (Dunnell and Feathers 1995). TL techniques are suf- ficiently refined that most pottery dating is consid- ered rather routine (Feathers 1997), and accurate dating has been demonstrated by several studies where independent dating evidence has been avail- able (e.g., Barnett 1999; Feathers and Rhode 1998; Kojo 1991). Nevertheless, because TL depends on a host of local variables, it is useful to demonstrate accuracy for any given case. Common systematic errors are caused by changes in the dose rate through time, uncertainties in the environmental (gamma) dose rate, changes in the moisture content through time, and anomalous fading. Dose rate is the amount of absorbed radiation per unit time, which is also equivalent to the radioactivity. Anomalous fading is unexpected loss of the potential luminescence sig- nal through time.
There is also the possibility that the event dated does not correspond with the archaeological event of interest. For example, the last heating could have occurred much more recently due to natural causes such as forest fires or a later cultural event that involved fire. A catastrophic event such as forest fire was unlikely to reset the clock in the Morris Site 1 project area, as demonstrated by more than 100 tree- ring specimens collected from standing trees and stumps; none showed evidence of being burned.
A more serious problem might be what we term "the old-pot effect." This refers to the fact that the TL signal most often represents the fing of pottery. Nelson (1991) indicates wide variation in use-life among different cultural groups, therefore, the date of firing as reflected by the TL date may predate the site by a significant margin; consequently, heirloom pieces may have been manufactured several gener- ations prior to the target event. Sherds were also functional, as ladles or scrapers for example, long after the ceramic vessel was broken. Such re-use of pottery was common in the Southwest; therefore, the old-pot effect is more likely to apply to pottery types that have longer use lives. In an ethnoarchaeologi- cal study of pottery function, Skibo (1994) found that the Kalinga people in the Philippines used cooking pots on a daily basis, but ceremonial vessels were used only periodically. In early Navajo contexts,
well-made, decorated pottery may have functioned for ceremonial or other nonutilitarian purposes and as a consequence may not have been subject to the hard daily use apparent from observable wear on
Table 2. TL Dates for the Morris Site 1 Project.
Site Number UW Lab No. Context Sample Type Reported Date" 11 196 254 Subsurface Sandstone 17271 t4361 B.C. 11196 253 Subsurface Sandstone 1535 t49 55836 25 1 Surface Jelnez BIW 1491 t41 55836 252 Surface Gobernador Polychrome 1753 -+ 43 79456 239 Subsurface Dinttah Gray 1597 + 66b 83529 147 Surface Dinttah Gray1734 + 32 83529 238 Subsurface Gobernador Polychrome 1685 t28b 105428 144 Surface Gobernador Polychrome 1766 + 30 105428 241 Subsurface Rio Grande matte paint 1612 + 39 105475 237b Subsurface DinCtah Gray 1489 t55 105475 237a Subsurface Dinttah Gray 1569 + 41 105479 145 Surface Sandstone 1665 + 50 105483 142Surface Gobernador Polychrome 1745 + 19 105530 143 Surface Gobernador Polychrome 1624 t32 105630 240 Subsurface DinCtah Gray 1615 t38 105929 242 Subsurface Gobernador Polychrome 1661 + 27 105930 244 Subsurface Dinttah Gray 1592 t42 105938 146 Surface DinCtah Gray 1718 t21 106168 246 Subsurface DinCtah Gray1726 + 24 106199 245 Subsurface Gobernador Polychrome 1676 t22b
106203 236 Surface Jemez BIW 1681 t26
110278 243 Subsurface Dinttah Gray 1649 t39
"Years A.D., one standard deviation, unless otherwise noted.
bDosimeter correction applied to date.
DinCtah Gray sherds. Decorated or imported pottery sured in the laboratory. Laboratory procedures and types were also more likely to be re-used as sherds. technical results have been reported elsewhere Under such circumstances the TL date may reflect (Feathers 2000b). an event much earlier than the target event. Cooking The age for one of the sandstone pieces, UW254, vessels, because of the stresses to which they are sub- predates any human habitation in the area. This sam- jected, have shorter use lives and may be more appro- ple was probably not heated sufficiently during the priate for dating. Neither postdepositional firing nor site occupation to reset the luminescence clock. The old-pot effects can be evaluated by luminescence age reflects a significant geological component to the methods alone. The comparison with tree-ring data signal and therefore will not be considered further. is then helpful in this regard. The precision of the dates, reflecting only analytical
TL samples were collected from both surface and errors, ranges from 6.9 to 18.8 percent, with a mean subsurface contexts (Table 2). Nineteen sherds and of 10.4 percent, which is about average for lumi- three burned sandstone pieces were obtained, mostly nescence dating of pottery. There was no tendency from middens because higher ceramic density for better precision among different types or between insured finding sherds of sufficient size for analy- surface and subsurface samples. The sherds from the sis3 On the other hand, middens may have more immediate surface, with one exception (UW25 I), complex radioactivity, and although sediments asso- tended to be somewhat younger than those a few cen- ciated with each sample were collected to assess the timeters below the surface, which would be expected environmental dose rate, direct measures of the on stratigraphic grounds. Dintitah Gray and Jemez radioactivity were deemed necessary. At four sites, Black-on-white tended to be older, while the poly- CaS0,:Dy dosimeters, encased in copper capsules, chromes tended to be younger with a few excep- were placed close to the sampling areas at about the tions. Two exceptions, a Jemez Black-on-white same depth as the recovered ceramics to provide fur- sample (UW236) and DinCtah Gray sample ther control on the external dose rate. These were in (UW246), suffered from anomalous fading, which place for one year and produced no significant dif- could explain their younger age. However, two other ference in external dose rate from the sediments mea- samples that indicated fading produced ages similar
Figure 3. Direct comparison of TL and tree-ring dates. Sites with better dating correspondence are shown on the left.
to others of their ceramic type. Excepting the anom- alous date (UW254), the sandstone pieces dated in the same age range as the gray and white wares, with comparable precision.
Tree-ring and TL Comparison
Three types of comparison are presented. Direct comparison refers to an evaluation of dating accu- racy within site contexts. In other words, the TL dates from any given site are compared with tree-ring results from the same context. The second compar- ison evaluates the results as aggregates of tree-ring- dated sites and TL results from the Morris Site 1 project area. This would represent the temporal dis- tribution of the early Navajo occupation of the pro- ject area. The final comparison evaluates the utility of certain ceramic types for dating purposes and the likely consequences of directly dating ceramic assemblages.
We use direct comparison of dates from same-site contexts to evaluate the correspondence of the TL results with respect to target events described by tree- ring dates. This analysis is based on the 12 sites from which the results of both techniques are available. Tree-ring dates for these sites are represented by cut- ting dates or estimated dates, derived by the termi- nal ring and sapwood analysis, and TL assays are represented by one- and two-standard-deviation ranges. The dates are charted (Figure 3) by site num- ber and ranked from left to right in decreasing order of correspondence to facilitate the comparison. Dat- ing correspondence is indicated when the tree-ring date occurs within the two-standard-deviation error range of the reported TL mean. A bridging argument is required to make sense of results where tree-ring dates occur outside TL ranges.
The precision of TL dates is an important con-
sideration in the direct comparison because poor pre- cision usually yields very large error ranges that increase the probability of direct correspondence with target dates. Thus, imprecision may provide accurate results because the larger date range offers a greater opportunity to hit the target date. Con- versely, better precision produces a narrower range that is statistically less likely to hit the target, but when these do correspond, the highest standard of precision and accuracy is achieved.
Earlier we characterized, as unacceptably impre- cise, the 400-year date range often provided by radio- carbon results in the protohistoric period. TL results should improve on that level of precision and show correspondence with the tree-ring target dates to be a viable alternative to radiocarbon dating. TL dates for all samples produced two-standard-deviation ranges that varied from 76-264 years with a mean range of 145 years. Consequently, the precision of the TL results shows considerable improvement compared to typical radiocarbon results for the period.
Correspondence between the TL range and tree-
ring date is indicated by 6 of 15 TL dates (see Fig-
ure 3). In these cases the correspondence between
the target tree-ring date and TL-dated event is
unequivocal, but it might be instructive to consider
the character of these samples. The sample materi-
als consisted of four Gobernador Polychrome sherds,
one Dinktah Gray sherd, and one piece of sandstone.
The sandstone sample demonstrates the poorest pre-
cision, indicated by the broadest date range of all TL
samples, and this likely factored into its correspon-
dence with the associated tree-ring date. Nonethe-
less, it is encouraging that the sandstone sample
provided accurate results, given the potential prob-
lem of insufficient heating. In the group of ceramic
sherd samples it is apparent that Gobernador Poly-
chrome provided more correspondent dates than
Dinitah Gray samples. This suggests that the accu-
racy of TL results may vary with ceramic type and
raises the issue of the relationship between dating
and sample character, which is explored later in the
Bridging arguments are necessary to evaluate the
nineTL dates that did not directly correspond to tree-
ring dates. Probably the most common reason for the
lack of correspondence is the inequality of the events
dated by the two techniques. This phenomenon can
be evaluated by closer inspection of sites with mul- tiple TL dates where both correspondence and lack of correspondence occur. Three sites, LA 55836, LA 83529, and LA 105428, meet this condition.
Two ceramic sherds from LA 55836 (Romine Canyon Ruin) were dated by the TL method. A sam- ple of Gobernador Polychrome provided a date range in the A.D. 1700s that corresponded well with the tree-ring estimate for the site. The noncorrespondent date in the 1ateA.D. 1400s was measured from a frag- ment of Jemez Black-on-white(UW 251). This ceramic type is known to have been widely traded in the Southwest, but originated from production centers in the Jemez area, located about 100 km south-southeast of the DinCtah (Harlow 1973). Pro- duction of Jemez Black-on-white cross dates with the tree-ring record between A.D. 1300-1700 (Bre- ternitz 1966). The painted designs of this decorated ware may have been favored by the Navajo for heir- loom pieces (Cleveland et al. 1999; Langenfeld
1999). Moreover, Jemez Black-on-white is quite stur- dily built and individual pieces could have a use life of considerable duration. This combination of attrib- utes suggests that the early TL date for the Jemez Black-on-white sherd may be attributed to the old- pot effect.
The situation at LA 105428 is similar to that just addressed for Romine Canyon Ruin. The later, cor- respondent date is derived from a piece of Gober- nador Polychrome, but the earlier date is from a fragment of untyped Rio Grande matte paint poly- chrome pottery. In this case, however, the early date is unexpected, because production of such pottery does not begin until ca. A.D. 1650 in the Rio Grande area (Harlow 1973). The most recent extreme of the TL date range is A.D. 1690, which does correspond with the earliest production of this pottery. Conse- quently, at two standard deviations the TL date could accurately date the production of this pottery. This result is still about 50 years older than the tree-ring target date, thus the relationship between site and dated event is best regarded as due to the old-pot effect.
Two TL dates from LA 83529 (Morris Site 1) show some overlap at one standard deviation and considerable overlap when the range is broadened to two standard deviations. The statistical similarity of TL ranges suggests that the same or similarly aged events are dated; however, only one of these dates showed correspondence with the tree-ring date. The character of the occupation at Morris Site 1 must be
considered in building the bridging argument for the noncorrespondent date. Morris Site 1 is a large and complex site that consists of numerous features, including a sweat lodge, hearths, and possible resi- dential structures (Brown 1993). The accretion of features suggests some duration of the occupation and noncutting tree-ring dates prior to 1749 might represent earlier use of the site. Therefore, it is pos- sible that the Morris 1 pueblito represents one of the last structures built at the site. In this case, the bridg- ing argument relies on physical data from the site, consideration of other tree-ring dates, and statistical similarity between multiple TL results, which com-
bined, imply that both TL dates are accurate.
There is little independent data that can be brought to bear on the six remaining TL dates that do not cor- respond with the target tree-ring dates. In most cases the difference between the TL range and tree-ring date is 30 years or less. Inequality of dating events may be argued for the LA 11 196 results. This site, like Morris Site 1, is extensive and contains a large and varied feature assemblage, which may be indica- tive of a lengthy occupation or perhaps multiple occu- pational components. The other five noncorrespondent TL dates may also be the product of dated-event inequality, but this cannot be resolved by closer examination of site data or tree-ring dates. Tree-ring dates at these sites tend to be cutting dates or clusters of dates indicating good confidence in a narrow range of possible dates for the target event.
Comparison of sample types indicates that four
of the six noncorrespondent samples are either Din&
tah Gray or sandstone, and two samples are Gober-
nador Polychrome. This is a virtual mirror image of
the distribution of sample types correspondent with
tree-ring results and may indicate some loss of accu-
racy for TL dates derived from DinCtah Gray, per-
haps due to the old-pot effect.
In summary,6of 15 TL samples unequivocally cor-
respond to tree-ring dates from same-site contexts.
Bridging arguments are invoked convincingly to
explain apparent discrepancies between dating and
target events in 3 of the 9 remaining cases. Conse-
quently, in 9 of 15 cases the correspondence between
the dating techniques is knowable and often quite
good. That leaves 6 sites where there is insufficient
data to build convincing bridging arguments, but
inequality of dating events is the prime suspect for the
date discrepencies. Even so, the differences between
the TL ranges and target dates are relatively small.
The precision and accuracy of the TL dates appear to be related in p@ to the type and character of the sample. On average the TL mean dates occurred within 79 years of corresponding tree-ring dates. Three of four TL mean dates that departed from tree- ring dates by greater than 100 years were assayed from sandstone or trade-ware pottery samples that may reflect old heating events. In contrast, the aver- age departure of then mean for locally made ceram- ics, Dinktah Gray, and Gobernador Polychrome is only 54 years. Moreover, Gobernador Polychrome samples appear to be more reliable than DinCtah Gray because the average departure from tree-ring dates is 44 years. The figure for Dinktah Gray is 71 years. Better accuracy for TL dates derived from Gobernador Polychrome may be due to the relatively narrow time frame, about 135 years, in which this pottery was produced. DinCtah Gray was produced for 350 years or more. Gobernador Polychome, therefore, may have the potential for producing dates
with better accuracy due to its shorter period of pro- duction. The accuracy of sandstone samples for dat- ing may suffer from a lack of consistent heating due to multiple heating events, whichmay affect the abil- ity to isolate the TL signal that relates to the event of interest. When sufficiently heated, the resultant dates from sandstone demonstrate dating resolution similar to DinCtah Gray samples. Finally, trade-ware pottery produced outside of the DinCtah area may be unreliable because production of these types was not closely linked with local site use. Therefore, better dating accuracy is associated with locally produced pottery.
The direct comparison shows that the TL dates are generally in the same range as tree-ring dates, but it is apparent that significant discrepancies can occur. In some cases the discrepancies can be explained if the respective events being dated are also shown to be separated in time-inequality of dating events. On a project-wide scale these differences might average out and provide an accurate representation of the time of occupation. This requires grouping TL and tree-ring dates into two data series and comparing the two on the same time scale. The construction of each data series was treated differently to accom- modate the differences in the way dates are reported.
The tree-ring dates are presented as a histogram with dated sites classed into bins of equal 20-year
lESWTree-ring Categories -+TL Categories
|0~||0||0 ~||0||0||0 $||0||0 $||0||0 ~||0||0 0 r~||0 n w|
|A.D. Date Category|
|(Center point of 20-year Interval)|
|Figure 4. Comparison of aggregated TL and tree-ring dates.|
c oo *
intervals. Histogram categories are designated by the center date within the interval. The raw site fre- quencies were converted to relative frequency in per- cent. Tree-ring dates could be treated in this fashion because each date represents a single value or in some cases a very narrow range of possible values.
TL dates represent a broader range of values indi- cated by the reported mean and one standard devi- ation; consequently, a statistical method was used to produce a comparable histogram. The TL dates were summarized using Kintigh's 14C routine (Kintigh 19941, which was originally designed to evaluate radiocarbon dates reported at one standard deviation. Instead of assuming each date as a single point in time, the routine treats each date as a nor- mally distributed probability about the mean. For any time interval, a given date will have some prob- ability of falling within that interval. For each time interval in the histogram, the probabilities for each date in that interval are summed, giving a total prob- ability for the number of dates for each interval. The histogram values are converted and reported as rel- ative frequencies. The resulting distribution indi- cates one or more modes that reasonably
demonstrate archaeological occupations (Kintigh 1990, 1994).
Both distributions are overlaid in Figure 4, so that the fit can be compared visually. The distribution of tree-ring dates is displayed as a series of vertical bars. To facilitate the comparison, the TL distribution is indicated by a series of points connected by lines and the points represent histogram categories designated by the center date for each 20-year interval.
The tree-ring distribution indicates three separate occupations in the Morris Site 1 project area between
- 1600 and A.D. 1940. The fxst occupation is indicated by only 5 percent of tree-ring dates in the
- 1620 category (A.D. 1610-1630). The second occupation appears to be longer, spanning four inter- vals between A.D. 1670 and A.D. 1750. This is fol- lowed by a long occupational hiatus spanning six intervals between A.D. 1750 and 1870. After A.D. 1870, historic wood harvesting indicates an early Lucero-phase use of the area.
The distribution of TL dates appears to be nega- tively skewed with a single mode evident in the A.D. 1700 category. If 100 percent of the distribution spanned by the TL curve is accepted, then the total
Percent TL Figure 5. TL-date distribution for DinCtah Gray samples.
possible range of the occupation extends fr0mA.D. 1360-1860. This range contains all variation given by the dates and is excessively broad for the purposes of this analysis. Similarly, a ninety-fifth percentile (two standard deviations) adjustment is too broad for comparative purposes. One standard deviation (sixty- sixth percentile) might be suitable for evaluating data from a single site, but with multiple sites represented, some temporal variation must be accepted. We use the seventy-fifth percentile here because it falls between one and two standard deviations. At the sev- enty-fifth percentile an occupation span of 200 years fr0mA.D. 1540-1740 is indicated. The single mode suggests that a single occupation is dated by the TL technique.
In comparison with the tree-ring results, the TL distribution does not reflect the historic occupation of the project area because there was no attempt to date historic remains with this technique. In the early Navajo period the correspondence of the TL mode with the strongest tree-ring mode is impressive. Both distributions support a significant early Navajo occu- pation between A.D. 1690 and A.D. 1750. The tree- ring data indicate the terminus of the occupation after A.D. 1750. The TL distribution indicates the terminus in the A.D. 1740 category (A.D. 1730-1749), just short of the A.D. 1750 boundary. The comparison indicates a great deal of correspon- dence between the TL and tree-ring results in this
The skewness of the TL curve prior to the A.D. 1680 category appears to be responsible for a lack of fit in the early part of the occupation. In order to replicate the tree-ring results, another mode should be evident in the TL distribution in the A.D. 1620 category. Such a mode is not indicated; however, a slight change in the slope of the TL distribution is evident after the A.D. 1620 category. This slight flat- tening of the curve indicates fewer summed TL results in the A.D. 1640 category and may represent the only evidence in the TL distribution for an ear- lier occupation.
|7 1||I||,'||I||:||75th Percentile||I|
|I :||Gobernador Polychrome TL-Date Distribution||I|
0 2 4 6 8 10 12 14 Percent TL
Figure 6. TL-date distribution for Gobernador Polychrome samples.
To evaluate the TL-dating distribution for individual pottery types, the l4 C program was used to produce separate histograms for dated Dinktah Gray and Gob- ernador Polychrome sherds (Figures 5 and 6). Rel- ative to DinCtah Gray the Gobernador Polychrome distribution is shifted to more recent times and is a strong contributor to the primary A.D. 1700 mode in the date distribution of all ceramics. The shift is likely due to the later period of Gobernador Polychrome production. Gobernador Polychrome date distribu- tion exhibits a single mode and a rather narrow range of A.D. 1640-1780 at the seventy-fifth percentile. The TL results show good fit with the expected dates
of A.D. 1640-1800 for this ceramic type (Brugge 1981; Hill 1995).
The distribution of DinCtah Gray dates shows a much broader range from theA.D. 1520s to theA.D. 1740s (seventy-fifth percentile) and is bimodal. One mode corresponds well with the Gobernador Poly- chrome distribution and the other mode appears to represent an earlier group of dates that may be largely responsible for the skewness of the overall distribu- tion (see Figure 4). At the seventy-fifth percentile the date distribution for Dinktah Gray fits well with the expected range for this ceramic type. Moreover, deconvoluting the overall TL distribution by ceramic types successfully reproduces the bimodality evi- dent in the tree-ring dates and indicates at least two Navajo occupations of the project area.
The agreement with known age distributions for these ceramics suggests that TL can be very reliable for dating ceramic production periods and may serve as an independent test for current ceramic cross-dat- ing sequences. TL dating may be useful for expand- ing ceramic cross dating to pottery types that are not currently keyed by reliable tree-ring dates. For dat- ing occupations at individual sites, reliability may be compromised to some degree by production peri- ods that extend longer than the site duration-the old-pot effect. Pottery types that had short periods of production will likely return more precise TL dates than pottery with long periods of production.
The evaluation of tree-ring and TL correspondence indicates good agreement considering the many potential sources of dating disparity that might be expected from two profoundly different dating tech- niques. In the direct comparison of dates from the same-site context, 6 of 15 (40 percent) TL samples col-responded with tree-ring dates. Bridging argu- ments provided logical explanations for lack of dat- ing correspondence for 3 other samples. The remaining 6 TL samples were found to be noncor- respondent compared to tree-ring dates. We believe that better dating col~espondence may yet be possi- ble given additional TL research and wider applica- tion and interpretation of TL dates.
TL dating consistently outperforms radiocarbon dating in both precision and accuracy for the proto- historic period, as is apparent when comparing the results here with those of Smilev and Ahlstrom (1997) and F~~~~~~~(1996). TL d,&g and tree-ring dating arebest viewed as techniques' each providing slightly different temporal informa- tion relative to the target event. Where tree-ring dates are available, they can also serve as a valuable inde- pendent check on theTL dates, which are inherently less precise. The TL technique is complementary because it has wider applicability for sample mate- rials such as ceramic sherds and burned rock, which are more abundant in archaeological contexts than the well-preserved wood necessaiy for tree-ring dat- ing. Aggregated TL data appear to provide excellent correspondence with tree-ring results and should be used with confidence where tree-ring dates are not
available. Bridging arguments may be necessary to account for variability in ceramic production dura- tion and the old-pot effect. Nonetheless, an aggre- gate of TL dates from a single site would significantly improve reliability of individual site dating.
We suspect that in different archaeological and environmental contexts the TL results will be more or less accurate than reported here. Consequently, TL and radiocarbon dating should be used together as part of a comprehensive site-dating strategy that will likely provide greater accuracy in archaeological site dating.
Aggregating the dates from multiple sites effec- tively averages the temporal differences between individual events and produces impressive corre- spondence in dating results. Close correspondence in both the modes and the overall distributions is apparent and may indicate the growth, fluctuations, cations for the Southwest's most commonly applied relative dating technique, ceramic cross dating, because it provides an accurate temporal profile of ceramic production. This may be used to calibrate ceramic cross dating in the absence of tree-ring dates. Finally, the suitability of TL dating for common archaeological materials and effectiveness for dat- ing surface artifacts make it a good choice for archae- ological survey where absolute dating is rarely used.
Acknowledgements. The Morris Site 1 project was funded in its entirety by Williams Field Services, Inc. Representing Williams Field Services at various times during the project were Larry Lauderback, Kathleen Reid, Paul Lehnnan, and Ken Clagett, who were most supportive of the archaeological research. All archaeological work for the project was conducted under Bureau of Land Management Cultural Resource Use Permit 10-8152 (1994-1996) and State of New Mexico Permit 95-036. The Navajo Nation Archaeology Department and Cultural Resources Management Consultants conducted
the project jointly. The office staff and field crews from both
organizations are commended for developing an atmosphere of
cooperation and support that has produced a quality product.
The Laboratory of Tree-ring Research and University of
Washington Ther~noluminescence Laboratory analyzed the
dating samples. Jeffrey Dean of the tree-ring lab joined us in
our first field season and lent his considerable expertise to the
problem of sampling culturally modified trees. Larry Vogler
and Jeffery Wharton provided comments on the original
research design. John Torres commented on portions of the
manuscript and assisted with troublesome graphical presenta-
tions. Lenora Tsosie drafted the base map in Figure 1. Kristin
Langenfeld provided technical editing of the final manuscript.
Clara Alonzo (Badajoz, Spain) and Lynda Valencia
(Farmington Schools Bilingual Program) prepared the Spanish
translation of the abstract. Finally, we would like to thank the
reviewers, Patrick Hogan, Stephen E. Nash, Janet Rafferty, and
two anonvmous reviewers for constructive comments that
greatly improved the manuscript; ultimately however, the
responsibility for errors of commission and omission rests
with the authors.
Aitken, M. J. 1985 Thennoluminescence Daring. Academic Press, London. Ayers, D. O., and L. S. Reed 1993 Res~~lts
of Archaeological Investigations at Two Dine'- tah Phase Sires (LA 72353 and LA 78481)near Smith Pass, RioArriba Count): NewMexico. TechnicalReportNo. 2710. Salmon Ruins Museum, Division of Conservation Archae- ology, Bloomfield, New Mexico.
Bailey, G., and R. G.Bailey 1986 A Histoiy of the Navajos: TheReselvarion Years. School of American Research Press, Santa Fe, New Mexico. Bailey, M. G. L. 1995 A Slice Through Time: Dendroclzronology and Preci-
peak occupation, and abandonment of the Mol~is sion Dating. B.T. Batsford, Ltd., London.
Site 1 project area duling the early Navajo period. Bannister, B., J. W. Hannah, and W. J. Robinson 1966 Tree-Ring Datesfrvm Arizona K: Puerco-Wide Ruin-
The direct TL dating of sherds may have impli-
Goodman, J. M. 1982 The Navajo Atlas. University of Oklahoma Press, Nor- man, Oklahoma. Graves, M.
1991 Estimating Ring-Loss on Tree-Ring Specimens from East-Central Arizona: Ilnplications for Prehistoric Growth at Grasshopper Ruin. Journal of Quantitative Anthropology 333-115.
Harlow, F. H. 1973 Matte-paint Pottery of the Tewa, Keres and Zuni Pueb- los. Museum of New Mexico Press, SantaFe, New Mexico. Haskell, J. L. 1987 Southern Athapaskan Migration: A.D. 200-1 750. Navajo Community College Press, Tsaile, Anzona. Hester, J. J. 1962 Early Navajo Migrations and Accztltztration in the Soztthwest. Museum of New Mexico Papers in Anthropol- ogy No. 6. Museum of New Mexico Press Santa Fe, New Mexico. Hill, D. V. 1995 A Brief Overview of the Navajo Presence in the Upper San Juan Drainage and South Colorado and their Ceramics. In Archaeological Pottery of Colorado: Ceramics Clues to the Prehistoric andPlatohistoric Lives of the State's Native Peoples, edited by R. H. Brunswig, Jr., B. Bradley, and S.
M. Chandler, pp. 98-128. CCPA Occasional Papers No. 2. Colorado Council of Professional Archaeologists, Denver, Colorado.
Hodge F. W., G. P. Hammond, and A. Rey
1945 Fray Alonso de Benavides' Revised Memorial of 1634. Coronado Cuarto Centenial Publications, 1540-1 940, vol- ume IV, edited by George P. Hammond. University of New Mexico Press, Albuquerque, New Mexico.
1989 DinBtah: A Reevaluation of Pre-Revolt Navajo Occu- pation in Northwest New Mexico. Journal of Anthropolog- ical Research 45(1):53-66.
Hogan, P., J. M. Elyea, and P. Eschnlan
1991 Ovewielv and Research Design for the Fruitland Coal Gas Development Area. University of New Mexico Office of Contract Archeology, Albuquerque, New Mexico.
John, E. A. H. 1975 Storms Brewed in Other Men's Worlds: The Con- fiontation of Indians, Spanish, and French in the Southwest,
1540-1 795. University of Oklahoma Press, Norman, Okla- homa. Kintigh, K. W.
1990 Intrasite Spatial Analysis: A Commentary on Major Methods. In Mathematics and Information Science in Archaeology: A Flexible Framework, edited by A. Voorrips. Studies in Modem Archaeology 3: 165-200. Holos, Bonn, West Germany.
1994 Tools for Quantitative Archaeolog)! Programs for Quan- titative Archaeology. Publisher? Tempe, Arizona. Kojo, Y. 1991 The Reliability of Thermoluminescence Dating: A Pilot Experiment. Geoarchaeology 6:367-374. Kotyk, E. M., and J. D. Cater
1998 Results of Data Recovery at Sites LA 85369 and LA 89898Located on the Phillips Petroleum Company Unit 296 Gathering System Lateral P-15 and the 30-5Gathering System No. 232 Well Tie Rio Arriba Count)r New Mexico. San Juan College Research Papers in Anthropology No. 9. San Juan College Cultural Resources Management Program, Farmington, New Mexico.
1999 Pottery as aMeasure of Change and Continuity in Early Navajo Households. Paper presented at the 64th Annual Meeting of the Society for American Archaeology, Chicago.
Marshall, M. P. 1995 A Chapter in Early Navajo History: Lnte Gobernador Phase Pueblito Sites of the Dine'tah District. University of New Mexico Office of Contract Archeology, Albuquerque, New Mexico. McNitt, F. 1972 Navajo Wars. University of New Mexico Press, Albu- querque, New Mexico. Nash, S. E. 1997 Archaeological Cutting Date Estimation and the Inter- pretation of EstimatedTree-Ring Dates. American Antiquity 62:260-272. Nelson, B. A. 1991 Ceramic Frequency and Use-life: A Highland Maya Case in Cross-cultural Perspective. In Ceramic Ethnoar- chaeology, edited by W. A. Longacre, pp. 162-181. Uni- versity of Arizona Press, Tucson, Arizona. Plog, S. 1980 Stylistic Variation in Prehistoric Ceramics: Design Analysis in the American Southwest. Cambridge University Press, New York. Reed, P. F., and L. S. Reed 1996 Reexamining Gobemador Polychrome: Toward a New Understanding of the Early Navajo Chronological Sequence inNorthwesternNew Mexico. In The Archaeology ofNavajo Origins, edited by R. H. Towner, pp. 83-108. University of Utah Press, Salt Lake City, Utah. Reeve, F. D. 1957 Seventeenth Century Navaho-Spanish Relations. The New Mexico Historical Review 32:36-52. 1958 Navaho-Spanish Wars 1680-1720. The New Mexico Historical Review 33:205-23 1. 1959 Navaho-Spanish Peace 1720's-1770's. The New Mex- ico Historical Review 34:940. Skibo, J. M. 1994 Pottery Function: A Use-Alteration Perspective. Plenum Press, New York. Slniley F. E., and R. V. N. Ahlstrom 1997 Archaeological Chronometry: Radiocarbon and Tree- ring Dating, Models andApplicationsfrvm BlackMesa, Ari- zona.Occasional Paper Number 16. Southern Illinois University at Carbondale Center for Archaeological Inves- tigations, Carbondale, Illinois. Stokes, M. A,, and T. L. Smiley 1968 An Introduction to Tree-ring Dating. University of Chicago Press, Chicago. Towner, R. H. 1997 The Dendrochronology of the Navajo Pueblitos ofDine'- tah. Ph.D. dissertation, Department of Anthropology, Uni- versity of Arizona, Tucson, Arizona. University Microfilms, Ann Arbor, Michigan. 1999 Dendrochronology in the Morris Site 1 Project Area. Ms. on file, Navajo Nation Archaeology Department, Farm- ington, New Mexico. Towner, R. H., and B. P. Johnson 1998 The San Raphael Survey: Reconstructing Eighteenth Centztly Navajo Population Dynamics in the Dine'tah using Archaeological and Dendrochronological Data. Arizona State Museum Archaeological Series No. 190. Arizona State Museum, University of Arizona Press, Tucson, Arizona. Towner, R. H., L. Sesler, and T. Hovezak 1998 Navajo Culturally Modified Trees in the DinBtah. In Dine' Bike'yah: Papers in Honor of David A. Brztgge, edited
by M. S. Duran and D. A. Kirkpatrick. pp 195-209. Papers of the Archaeological Society of New Mexico No. 24. The Archaeological Society of New Mexico,Albuquerque, New Mexico.
Wharton, J. T., D. D. Dykeman, and P. F. Reed 1996 Small Site Archaeology in the Upper Sun Juan Basin: Investigations at Archaic, Anasazi, and Navajo Sites in Northwestern New Mexico, Fruitland Data Recovery Series NO, 1, ~~~~j~ &tion papers in~nkopology N~,
32,~~~~j~ Nation Archaeology Department, Window Rock, Arizona.
York, F., and J. Winter
1988 Report on Ethnographic Study and Archaeological Review ofProposed Coal Lease Tracts in Northwestern New Mexico. University of New Mexico Office of ContractArche- ology, Albuquerque, New Mexico.
- The terms "accuracy," "precision," and "resolution" are used to evaluate archaeometric dating techniques and are defined here for the purposes of consistent usage in this report. Accuracy is a measure of systematic error and refers to the closeness of our archaeolnetric estimate to the true value. Precision is a measure of random error and refers to how often we come up with the same answer on repeated tries or measurements. Resolution is closely related to preci- sion and refers to the shortest interval of time by which two assays can be distinguished. In effect, the most useful archaeolnetric dates should reflect the true date of an event (accuracy) within a narrow time frame (precision); the tech- nical ability of an archaeometric procedure to provide such results is regarded as resolution.
- The Bureau of Land Management-Farmington Field Office administers the Fruitland data recovery under a Memorandum of Agreement with the New Mexico State Historic Preservation Office. Archaeological data recovery
conducted under the Fruitland Project consists of many smaller projects sponsored by energy companies that produce natural gas from the Fruitland geologic formation in north- western New Mexico. Multiple archaeological firms are involved in the data recovery effort, which is guided by a sin- gle research design entitled Overview and Research Design
for the Fruitland Gas Area et al. 199 1). Guidelines for archaeological methods appropriate for research-design issues were prepared by the Farmington Resource Area Cultural Advisory Group (1991) and pub- lished in a document entitled Data Comparability Guidelines for Fruitland Coal Gas Gathering System Data Recovery. In addition, individual data-recovery plans are prepared for each project conducted under the research design. The Morris Site 1 project was implemented by a data-recovery plan prepared by the Navajo Nation Archaeology Department; the plan was entitled The Morris Site 1 Early Navajo Land Use Study: An Alternative Data Recovery Plan for Sites LA 11196, LA 83529, and LA 88766 in Williams Field Services Unit 29-7 Gas Gathering System, Rio Arriba County, New Mexico (Dykeman and Wharton 1994). The Navajo Nation Archaeology Department and Cultural Resources Management Consultants of Farmington, New Mexico conducted fieldwork on the project between 1994 and 1996. The final published report on the Morris Site 1 pro- ject is expected in the fall of 2001.
3. Samples submitted for TL assay are a minimum of 3 cm in diameter and .5 cm in thickness. Exterior portions of the samples are removed before the assay is performed to reduce the influence of environmental factors.
Received April 26, 2001; Revised July 25, 2001; Accepted
July 30, 2001.