Introduction
The Caribbean is an underrepresented region for
studies into the compositional nature of ancient
material culture (but see Carini 1991; Crane 1993;
Hofman and Bright 2004; Knippenberg 2006;
Padilla et al. 2006). Compositional analyses of ar-
tifacts can provide insights into issues of produc-
tion, use, exchange, consumption and diagenesis.
With large representative data sets from archaeo-
logical sites and resources, issues of ancient pop-
ulation movements into the region, the develop-
ment of exchange networks within and among
various island archipelagoes, and local adapta-
tions and prehistoric developments can be ad-
dressed.
This compositional study investigates the
chemical compositions of ceramic assemblages
from five archaeological sites on the Caribbean is-
land of Antigua, Lesser Antilles, for an under-
standing of compositional changes through time
and insights into ceramic production. The 102 ce-
ramic specimens were collected from the follow-
ing five sites (Table 1): Indian Creek (PA-04),
Royall’s (JO-11), Mill Reef (PH-01), Winthorpe’s
East (GE-01) and Muddy Bay (PH-14). Previ-
ously, we investigated ceramic sherds samples
from the Indian Creek site (Descantes et al. 2007).
In 1973, the late Professor Irving Rouse of Yale
University and the Antigua Archaeological Soci-
ety excavated several middens at the Indian Creek
site (Rouse and Morse 1999). Yale’s Peabody Mu-
seum of Natural History permitted us to analyze
a sample of their extensive collection. The ceramic
specimens from the remaining four sites were all
excavated in the 1990s and generously supplied by
Dr. Reginald Murphy, Curator and Archaeologist
of the Antigua and Barbuda national parks.
Antigua has an island area of 284 km2 and is
situated in the Guadeloupe Passage of the Lesser
Antilles in the Caribbean Sea (Figure 1). The In-
dian Creek site, with a ceramic assemblage from
Neutron Activation Analysis of Ceramics 
from Five Archaeological Sites in Antigua, West Indies
Christophe Descantes,1 Robert J. Speakman2 and Michael D. Glascock3
1Archaeological Research Facility, 2251 College Building, 
University of California, Berkeley CA 94720-1076 USA
—email: cdescantes@berkeley.edu
2Museum Conservation Institute, Smithsonian Institution, 
Museum Support Center, 4210 Silver Hill Road, Suitland MD 20746 USA
3Archaeometry Laboratory, Research Reactor Center, 
University of Missouri–Columbia, 1513 Research Drive, Columbia MO 65211 USA
Abstract
Ceramics of multiple styles from five archaeological sites on Antigua were analyzed at the Uni-
versity of Missouri Research Reactor Center to obtain a better understanding of ceramic compo-
sitional variability within Antigua. Bulk chemical characterizations of pottery were determined by
neutron activation analysis. Findings suggest that all of the ceramics in our sample were made
from locally available, naturally tempered clays, and that there is compositional continuity
through time despite the different stylistic and locational attributes of the sherds.
Keywords
Caribbean, provenance studies, Antigua, ceramics, NAA.
Bulletin of the Peabody Museum of Natural History 50(1):147–155, April 2009.
© 2009 Peabody Museum of Natural History, Yale University. All rights reserved. • http://www.peabody.yale.edu
all three Ceramic Age series, lies between two in-
lets, English Harbour to its west and Marmora Bay
to its east. One kilometer inland from the north-
ern coast, ceramic sherds from the site of Royall’s
are diagnostic of the early Saladoid (Murphy
1999:144). The Mill Reef site, which has been the
focus of much research, is located on the east coast
of Antigua. The 10 sherd specimens from the Mill
Reef site are recognized as part of the Terminal
Saladoid on Antigua (Murphy 1999:205). The ce-
ramic assemblage of the Winthorpe’s East site, sit-
uated in the waterfront of Winthorpe’s Bay, is di-
agnostically Post-Saladoid (Murphy 1999:69).
Situated on the east coast of Antigua, the Muddy
Bay site ceramics are also identified as Post-Sal-
adoid (Murphy 1999:250).
The island is characterized by three distinc-
tive geological regions (Figure 2): the Basal Vol-
canic Suite, the Central Plain Group and the
Antigua Formation (Multer et al. 1986:2–7). The
Basal Volcanic Suite region, located in the south-
western reaches of the island, is composed of Late
Oligocene agglomerates, lavas, tuffs and intru-
sions. The Indian Creek site is situated on sedi-
ments of volcanic origin; the region may also
contain occasional lenses of shell- and coral-frag-
ment limestone, deposited on the sea floor dur-
ing volcanic eruption intervals (Tomblin 2005:13).
The Central Plain Group includes mixed terres-
trial and marine sediments, whereas the Antigua
Formation comprises limestone uplands. All sites,
except the Indian Creek site, are on the Antigua
Formation, which is made up of limestone layers
laid down between 30 and 23 million years ago
(Tomblin 2005:15). The limestone uplands con-
tain occasional layers of compacted volcanic dust
from renewed small-scale eruptions, as well as
Figure . Antigua among the islands of the Caribbean.
Bulletin of the Peabody Museum of Natural History 50(1) • April 2009148
Figure . Five site locations and the three geological
provinces of Antigua. Legend: GE-01, Winthrop’s
East; JO-11, Royall’s; PA-04, Indian Creek; PH-01,
Mill Reef; PH-14, Muddy Bay.
minor layers of clays, sands and gravels from the
eroding flanks of volcanic peaks (Tomblin 2005:
15).
The five assemblages (Figure 3) under study
span the Ceramic Age of Antigua, that is, the Sal-
adoid (AD 71 to 600), Terminal Saladoid (AD 600
to 900), and Post-Saladoid (AD 900 to 1500).
Compared with the rest of the Caribbean, peoples
with ceramic technology arrived late on Antigua.
Only ceramics from the poorly understood Free-
man’s Bay Phase (AD 1200 to 1500), which have
only been identified at the site of Freeman’s Bay
(PA-01) in southeast Antigua, are not represented
in this study. The first pottery-making style found
on Antigua, the Saladoid tradition, originated
from the Orinoco region of present-day Vene-
zuela (Rouse and Morse 1999). The maritime
Amerindians lived in large sedentary villages and
cultivated root crops, practicing what Petersen
(1997:119) has called a tropical forest subsistence
economy. Plain pottery and two decorated wares
characterize the Indian Creek complex (Saladoid)
at the Indian Creek site. Rouse and Morse (1999)
described the first decorated ware as a relatively
thin white-on-red ware (wor), which has fine
particles of grit temper, many vessel shapes and a
combination of modeling, incision or punctation.
The second ware is a  zoned-incised-crosshatched
ware (zic), lighter in color, thin-walled, finely
tempered with grit or sand, and only found in
bowl vessel shapes. Most are unpainted; many are
red-slipped like their wor counterparts (Rouse
and Morse 1999:29). Thick-walled disc-shaped
griddles and cylinders comprise the other ceramic
artifacts of the complex. The diagnostic ceramic
traits of Saladoid wares at Royall’s also include
wor, polychrome and overall white painting, zic,
D-shaped strap handles, rounded handles with
short cylindrical nobs, tabular lugs, and flanged
rims that are at times incised or decorated with
raised or button-like nubbins (Murphy 1999:148).
The Mill Reef site typifies the divergence from
the traditional Saladoid to the Terminal Saladoid.
The second chronological series of ceramic and
cultural development on Antigua is the Terminal
Saladoid (Murphy 1999). The Mill Reef complex
is characterized by ceramic style changes and an
increase in settlements on Antigua. Mill Reef pot-
tery is characterized as thicker, cruder and lighter
in surface color than the early Saladoid pottery
(Murphy 1999:206). wor painted vessels are pres-
ent, but there is a noticeable deterioration in qual-
ity, sophistication, artistry and technological abil-
ity. No zic sherds, decorated rims, adornos,
modelling, D-shaped strap handles, round button-
like nubbins or polychrome painted vessels were
recovered. Finely incised labial flanges and tabu-
lar lugs are also absent (Murphy 1999:206). Simi-
larly, Rouse and Morse discovered that only wor
was found in the Mill Reef Complex at the Indian
Creek site, and the diversity of vessel shapes and
the quality of construction decreases. Smooth sur-
face finishes, elongated vertical ridges modeled
into zoomorphic heads, and simple rectilinear de-
signs of mostly isolated groups of parallel white
lines are introduced in this complex (Rouse and
Morse 1999:36–37). Red slip washes still dominate
in the ceramic assemblage of the complex, as do
the presence of ceramic cylinders and griddles.
The ceramics of the Post-Saladoid have been
described as relatively hard, thick-walled and poor
in quality, according to Rouse and Morse (1999:
39) when characterizing the ceramics from the
Mamora Bay complex at the Indian Creek site.
Other ceramic characteristics of this complex at
Indian Creek include the persistence of bird’s head
lugs, plain strap or rod handles, and semicircular
or rectangular tabs, in addition to large particles of
Neutron Activation Analysis of Ceramics • Descantes et al. 149
Antiguan sites
Ceramic Age Indian Creek Royall’s Mill Reef Winthorpe’s East Muddy Bay
culture series PA-04 JO-11 PH-01 GE-01 PH-14
Saladoid 24 10 0 0 0
Terminal Saladoid 19 0 10 0 0
Post-Saladoid 19 0 0 10 10
Table . Cultural and site provenience of ceramic specimens from five archaeological sites on Antigua. 
grit. Of the decoration, scratching, scoring and red
slipping remain common, while incised lines
become broader and deeper. Murphy (1999) finds
diagnostic attributes of this Post-Saladoid to be
broad-line, curvilinear incisions, overall red slip,
thickened and folded wedge-shaped rims, scratched
surfaces and, although not common, crudely ap-
plied white paint. The ceramic changes are accom-
panied by increases in settlement numbers and a
change in subsistence strategy to greater use of ma-
rine resources (Murphy 1999).
Analytical Methods
The compositional data are used to address issues
of compositional variability, production and use,
as well as changes through time. Neutron activa-
tion analysis (NAA) was applied to sherd speci-
mens from five Antiguan sites to investigate the
compositional variation of the ceramic pastes.
Representative ceramic specimens were chosen
from the five Antiguan assemblages for the com-
positional analyses. NAA is a precise and accurate
analytical technique for determining elemental
abundances of materials. In brief, NAA involves
activating the nuclei of elements in the sample by
bombarding them with neutrons generated from
a nuclear reactor. Once activated, radioactive nu-
clei emit gamma rays with characteristic ener-
gies. Detectors count the gamma radiation and
measure their energies to determine the concen-
trations of specific elements. Statistical algorithms
are then applied to the chemical data to identify
compositional groups and to calculate the prob-
abilities that an artifact originates from a partic-
ular source. The automated gamma-ray counting
system at the University of Missouri Research
Reactor Center (MURR) accumulates and stores
spectra, performs a peak search and a pulse pile-
up correction, and calculates elemental abun-
dances correcting for decay time, spectral inter-
ferences and sample weight. Abundances are then
calculated in relation to National Institute for
Standards and Technology standards (e.g.,
SRM1633a), which are irradiated at the same time
as the unknown samples. Neff ’s (2000) gauss
routines are used to reduce the data, which in-
cludes substituting missing data, generating bi-
variate plots and biplots, calculating principal
components, and calculating probabilities of
group membership based on Mahalanobis dis-
tances. The compositional data and accompany-
ing contextual data can be accessed online from
the MURR Archaeometry Laboratory (MURR
2008).
Results
Before identifying groups in the chemical data, we
applied a correction to the sherds with calcium
concentrations greater than 1% (see Steponaitis et
al. 1996; Cogswell et al. 1998). The calcium cor-
rection, which did not have an appreciable effect
on the group structure, was not used in a previous
NAA of the Indian Creek site ceramic specimens
(see Descantes et al. 2007). The high calcium
abundances are most likely a result of either the
original clay sources or a calcium-rich temper. As
in previous treatments of ceramics with enriched
calcium concentrations, we eliminated the ele-
mental concentrations of calcium and strontium.
Nickel and uranium concentrations of the data
were also eliminated because of their low detection
limits. Twenty-nine elemental abundances were
used to determine group membership.
A four-group structure was identified in the
Antiguan ceramic specimens: Group 1 (n=6),
Group 2 (n=4), Group 3 (n=74) and Group 4
(n=4). The compositional groups can be graphi-
cally represented in principal component space
(Figure 4) and in elemental space (Figure 5). Sta-
tistical tests based on Mahalanobis distances sup-
Bulletin of the Peabody Museum of Natural History 50(1) • April 2009150
Figure . Ceramic sherd samples. A. Indian Creek site (CANT052), Saladoid style. B. Royall’s site (CANT064),
Saladoid style. C. Mill Reef site (CANT100), Terminal Saladoid style. D. Winthorpe’s East site (CANT074),
Post-Saladoid style. E. Muddy Bay site (CANT088), Post-Saladoid style.
Figure . Variance–covariance matrix PCA biplot of principal components 1 and 3 showing the four com-
positional groups identified in the Antiguan ceramic sample. Ellipses represent 90% confidence level for mem-
bership in the groups. Unassigned specimens (+) are not labeled.
Figure . Bivariate plot of base-10 logged chromium and iron concentrations showing the four compositional
groups identified in the Antiguan ceramic sample. Ellipses represent 90% confidence level for membership in
the groups. Unassigned samples (+) are not labeled.
Neutron Activation Analysis of Ceramics • Descantes et al. 151
port the graphical representation of the group
structure. Nine principal components (89% of the
variance) and a cut-off probability of 1% were
used to refine further the membership of Group 3.
The small membership sizes of the other identi-
fied compositional groups precluded robust sta-
tistical tests, leaving us to test statistically the
probability of the other members having mem-
bership in Group 3. Fourteen specimens (13.7%)
could not be assigned to any of the identified
compositional groups, but five of the unclassified
specimens (4.9%) resemble statistically the large
and chemically diverse Group 3 (see Figures 4 and
5). The plotted elemental vectors and the princi-
pal component data show which elements con-
tributed to the differentiation of the four groups
(see Figure 4). Group 1 is distinguished from the
other three compositional groups by having richer
concentrations of chromium, sodium, cerium and
hafnium. Compositional Group 4 tends to have a
higher concentration of vanadium than the other
groups. Groups 2 and 3, 76% of the specimens, are
intermediate in elemental concentrations relative
to the other compositional groups.
At this preliminary stage there is little com-
positional differentiation between the ceramics
found at the five sites (Figure 6, Table 2). Compo-
sitional Group 3 ceramics are found in all five sites
and Group 4 ceramics are only found on the sites
on the limestone-layered Antiguan Formation.
The Royall’s site only has Group 3 ceramics.
Group 2 compositional ceramics are only found
at the Indian Creek site. When examining the
compositional variation of the ceramics among
the three chronological culture series, all the
chemical compositional groups are found in the
three Ceramic Age series, except for Group 2,
which is only identified in the early Saladoid at the
Indian Creek site (Table 3, Figure 7). Future com-
positional studies on a larger representative sam-
ple may provide more compositional diversity and
definitive patterns of the Antiguan ceramic com-
positional space.
Discussion
A four-group structure based on chemical com-
position is identified in the Antiguan ceramic
sample. There does not seem to be strong patterns
of correlation between the chemical composition
Bulletin of the Peabody Museum of Natural History 50(1) • April 2009152
Figure . Variance–covariance matrix PCA plot of principal components 1 and 2 displaying the five Antiguan
site groups. Ellipses represent 90% confidence level for membership in the groups.
of the sherd specimens and their particular site
provenience and chronological style (see Tables 2
and 3, Figures 6 and 7). Despite small sample sizes
at four of the sites, Group 3, the largest and most
diverse compositional group in the sample, pre-
dominates in all five sites and all three chrono-
logical culture series. However, Group 2 ceramics
are only found during the Saladoid and at the In-
dian Creek site, and Group 4 ceramics are not
found in the Saladoid or at the Indian Creek site.
Throughout the temporal culture series at the five
Antiguan sites, the bulk of the ceramics from our
sample were made from similar clay source or ce-
ramic recipes, or both. Despite the ceramic and
sociocultural changes during this 1,500 year pe-
riod (see Keegan 2000), ceramic recipes seem to
have remained unchanged.
At present the data do not imply that the
chemical composition of ceramic recipes differed
greatly between sites or varied drastically over
time. However, the earlier compositional analysis
of ceramics from only the Indian Creek site, which
included potsherds from all three chronological
culture series, was interpreted as having more
compositional heterogeneity in the early Saladoid
pottery (Descantes et al. 2004, 2007). This may
have been linked to a standardization of ceramic
production or a gradual decline in the ceramic
technologies through time, as has been suggested
by Murphy (1999:313), Rouse and Morse (1999)
and others. The ceramic sample in this study did
not allow for a diachronic investigation of ceramic
compositional change at the other four sites.
We hypothesize that all four identified groups
were locally produced, given the geologic charac-
teristics of Antigua. The petrographic analysis
conducted by David Hill (2005, 2006) links the ge-
ology of Antigua with the analyzed ceramic spec-
imens. Hill suggests that the inclusions in the ce-
ramic pastes are natural, as opposed to added
temper. The ceramic potsherds that contain ex-
clusively volcanic rock fragments are more likely
to derive from the Basal Volcanic Suite in the
southwestern region of Antigua, whereas pots
containing both limestone and volcanic rock frag-
ments are made with clays from the Central Plain.
Compositional Group 1 ceramics seem to have
been made with clays deriving from the South-
west Volcanic District, whereas clays used for
compositional Group 2 ceramics derive from the
Central Plain District. The ceramic specimens in
Group 3 are composed of diverse varied miner-
alogical suites of volcanic rock and limestone frag-
ments. None of the Group 4 specimens received a
mineralogical examination. A raw material study
of Antiguan clays would permit us to conclude
with certainty whether the inclusions are indeed
natural and to tie the compositional groups to clay
source areas on the ground. The analysis of more
ceramics will allow us to identify subgroups
Neutron Activation Analysis of Ceramics • Descantes et al. 153
Antiguan sites
Chemical Indian Creek Royall’s Mill Reef Winthorpe’s East Muddy Bay
groups PA-04 JO-11 PH-01 GE-01 PH-14
Group 1 5 0 1 0 0
Group 2 4 0 0 0 0
Group 3 41 9 8 9 7
Group 4 0 0 1 1 2
Unassigned 12 1 0 0 1
Table . Site provenience and compositional group affiliation of ceramic specimens from five archaeological
sites on Antigua.
Ceramic Age culture series
Chemical Terminal Post-
groups Saladoid Saladoid Saladoid
Group 1 3 2 1
Group 2 4 0 0
Group 3 24 21 29
Group 4 0 1 3
Unassigned 3 5 6
Table . Compositional group affiliation and culture
series of Antiguan ceramic specimens.
within this large and diverse group.
Throughout the three temporal culture series
at the five Antiguan sites, the bulk of the ceramics
from our sample were made from similar clay
sources and ceramic recipes. This core group is
referred to as Compositional Group 3. The large
and diverse compositional Group 3 includes wor,
zic, plain and incised wares obtained from all the
Ceramic Age culture series. 
Conclusions
This preliminary study into the bulk chemical
compositional variability of prehistoric Antiguan
ceramics using NAA has shown that at least a
four-group structure can be identified. The core
compositional Group 3 was found to predominate
in all of the Ceramic Age culture series and the five
archaeological sites. Throughout the Antigua’s Ce-
ramic Age, most ceramics from our sample were
made from similar Antiguan clay sources and ce-
ramic recipes. New insights about past ceramic
production behaviors on Antigua will be gained
when future research includes a detailed study of
raw clays from Antigua to tie the compositional
groups to clay source areas. Finally, a larger sam-
ple of sherds from geographically diverse archae-
ological sites spanning the entire Ceramic Age
will allow for a more diachronic perspective into
the ceramic behaviors of Antiguan potters.
Acknowledgments
We thank Reginald Murphy and the late Ben
Rouse for access to their Antiguan ceramic col-
lections. Frank Hole, Birgit Faber Morse and Roger
Colten at Yale University’s Peabody Museum of
Natural History are thanked for their assistance.
Operating support for the MURR Archaeometry
Laboratory was provided by the National Science
Foundation (BCS-0504015). This research was
also supported in part by a grant from the US De-
partment of Energy Office of Nuclear Energy, Sci-
ence and Technology Award No. DE-FG07-
03ID14531 to the Midwest Nuclear Science and
Engineering Consortium under the Innovations in
Nuclear Infrastructure and Education program.
Received 30 May 2008; revised and accepted 25
December 2008.
Bulletin of the Peabody Museum of Natural History 50(1) • April 2009154
Figure . Variance–covariance matrix PCA plot of principal components 1 and 3 showing the three Antiguan
chronological culture series. Ellipses represent 90% confidence level for membership in the groups.
Literature Cited
[MURR] Archaeometry Laboratory, MURR. 2008. The
Archaeometry Laboratory at the University of Missouri Re-
search Reactor [internet]. c1996–2007. [updated 2008 Nov
21]. Columbia, MO: University of Missouri–Columbia.
Available from: http://archaeometry.missouri.edu/
Carini, S. P. 1991. Compositional Analysis of West Indian
Saladoid Ceramics and their Relevance to Puerto Rican
Prehistory [dissertation]. Storrs-Mansfield, CT: University
of Connecticut, Department of Anthropology. 187 pp.
Available from: ProQuest Dissertations and Theses: Full
Text [online database]; http://www.proquest.com/ (publi-
cation no. AAT 9227575).
Cogswell, J., H. Neff and M. D. Glascock. 1998. Analy-
sis of shell-tempered pottery replicates: implications for
provenance studies. American Antiquity 63(1):63–72.
Crane, B. D. 1993. Colono Ware and Criollo Ware Pottery
from Charleston, South Carolina and San Juan, Puerto
Rico in Comparative Perspective [dissertation]. Philadel-
phia: University of Pennsylvania, Department of Ameri-
can Civilization. 388 pp. Available from: ProQuest
Dissertations and Theses: Full Text [online database];
http://www.proquest.com/ (publication no. AAT 9331769).
Descantes, C., R. J. Speakman and M. D. Glascock. 2004.
Neutron Activation Analysis of Ceramics from the Indian
Creek Site, Antigua: Preliminary Results [manuscript].
Available from: Archaeometry Laboratory, University of
Missouri Research Reactor (MURR), University of Missouri,
Columbia, MO. 15 pp.
Descantes, C., R. J. Speakman, M. D. Glascock and D. V.
Hill. 2007. Chemical and mineralogical analyses of ceram-
ics from the Indian Creek site, Antigua: preliminary results.
In: B. Reid, ed. Proceedings of the 21st Congress of the In-
ternational Association for Caribbean Archaeology, Vol-
ume 1. St. Augustine, Trinidad and Tobago; 2005 July 24–30.
St. Augustine, Trinidad and Tobago: School of Continuing
Studies, University of the West Indies. pp. 355–361.
Hill, D. V. 2005. Petrographic Analysis of Ceramics from An-
tigua [manuscript]. Available from: Archaeometry Labora-
tory, University of Missouri Research Reactor (MURR),
University of Missouri, Columbia, MO. 10 pp.
— 2006. Petrographic Analysis of Four Ceramic Sherds from
Antigua [manuscript]. Available from: Archaeometry Lab-
oratory, University of Missouri Research Reactor (MURR),
University of Missouri, Columbia, MO. 4 pp.
Hofman, C. L. and A. J. Bright. 2004. From Suazoid to Folk
Pottery: Pottery Manufacturing Traditions in a Changing
Social and Cultural Environment on St. Lucia. New West In-
dian Guide 78(1–2):5–35.
Keegan, W. F. 2000. West Indian archaeology. 3. Ceramic Age.
Journal of Archaeological Research 8(2):135–167.
Knippenberg, S. 2007. Stone Artefact Production and Ex-
change among the Northern Lesser Antilles. Leiden:
Leiden University Press. 382 pp. Available from: Institu-
tional Repository Leiden University [dissertation]; http://
hdl.handle.net/1887/4433
Multer, H. G., M. P. Weiss and D. V. Nicholson. 1986. An-
tigua: Reefs, Rocks and Highroads of History. St. John’s,
Antigua, WI: Leeward Islands Science Associates. 107 pp.
(Contribution 1.)
Murphy, A. R. 1999. The Prehistory of Antigua, Ceramic
Age: Subsistence, Settlement, Culture and Adaptation
within an Insular Environment [dissertation]. Calgary:
University of Calgary. 350 pp. Available from: ProQuest
Dissertations and Theses: Full Text [online database]; http://
www.proquest.com/ (publication no. AAT NQ47905).
Neff, H. 2000. Neutron activation analysis for provenance de-
termination in archaeology. In: E. Ciliberto and G. Spoto,
eds. Modern Analytical Methods in Art and Archaeology.
New York: Wiley. pp. 81–134. (Chemical Analysis 155.)
Padilla, R., P. Van Espen and P. P. Godo Torres. 2006. The
suitability of XRF analysis for compositional classification
of archaeological ceramic fabric: a comparison with a previ-
ous NAA study. Analytica Chimica Acta 558:283–289.
Petersen, J. B. 1997. Taino, Island Carib, and prehistoric
Amerindian economies in the West Indies: tropical forest
adaptations to island environments. In: S. M. Wilson, ed.
The Indigenous People of the Caribbean. Gainesville: Uni-
versity Press of Florida. pp. 118–130.
Rouse, I. and B. F. Morse. 1999. Excavations at the Indian
Creek Site, Antigua, West Indies. New Haven: Yale Univer-
sity, Department of Anthropology and Peabody Museum of
Natural History. 70 pp. (Yale University Publications in An-
thropology 82.)
Steponaitis, V. P., M. J. Blackman and H. Neff. 1996.
Large-scale patterns in the chemical composition of Missis-
sippian pottery. American Antiquity 61(3):555–572.
Tomblin, J. 2005. The Geology of Antigua, Barbuda and Re-
donda: Rocks, Minerals, and Fossils, including Two Field
Trip Guides. St. John’s, Antigua: Sun Printing and Publish-
ing. 24 pp.
Neutron Activation Analysis of Ceramics • Descantes et al. 155