lable at ScienceDirect
Journal of Archaeological Science 81 (2017) 13e22Contents lists avaiJournal of Archaeological Science
journal homepage: http : / /www.elsevier .com/locate/ jasThe Anoka, Minnesota iron meteorite as parent to Hopewell
meteoritic metal beads from Havana, Illinois
Timothy J. McCoy a, *, Amy E. Marquardt a, 1, John T. Wasson b, c, Richard D. Ash d,
Edward P. Vicenzi a, e
a Dept. of Mineral Sciences, National Museum of Natural History, 10th and Constitution Aves NW, Smithsonian Institution, Washington, DC 20560-0119, USA
b Dept. of Earth, Planetary and Space Sciences, University of California, Los Angeles, CA 90095-1567, USA
c Dept. of Chemistry and Biochemistry, University of California, Los Angeles, CA 90095-1567, USA
d Dept. of Geology, University of Maryland, College Park, MD 20742, USA
e Museum Conservation Institute, Smithsonian Institution, 4210 Silver Hill Road, Suitland, MD 20746, USAa r t i c l e i n f o
Article history:
Received 21 July 2016
Received in revised form
27 January 2017
Accepted 12 March 2017
Available online 22 March 2017
Keywords:
Hopewell
Meteorite
Trade
Manufacture
Source material* Corresponding author.
E-mail addresses: mccoyt@si.edu (T.J. McCoy), A
(A.E. Marquardt).
1 Present address: Department of Chemistry and
Colorado at Boulder, 215 UCB, Boulder, CO 80309-021
http://dx.doi.org/10.1016/j.jas.2017.03.003
0305-4403/Published by Elsevier Ltd.a b s t r a c t
Although rare among Hopewell horizon artifacts, meteoritic metal represents the most exotic raw ma-
terial used during the Middle Woodland period in Eastern North America. We demonstrate that Hope-
well meteoritic beads recovered from Havana, Illinois can be linked to the Anoka, Minnesota, iron, which
fell as a shower of irons across the Mississippi River. The similarity in major, minor and trace element
chemistry between Anoka and Havana, the presence of micrometer-sized inclusions of gamma iron in
kamacite in both, and the obvious connection via the Mississippi and Illinois Rivers between Anoka and
Havana point to the production of the Havana beads from a mass of the Anoka iron. Experiments strongly
support the manufacture of the beads via fragmentation of schreibersite inclusions to liberate small
pieces of metal. Repeated cycles of heating to temperatures of 600e700 
C followed by cold-working
produced flattened metal sheets. These sheets were subsequently rolled to make the Havana beads.
Recovery of the iron mass of Anoka that was used to make the beads likely occurred by local populations
who were part of the Trempeleau Hopewell center, with exchange bringing it to the Havana Hopewell
center, where the beads were manufactured.
Published by Elsevier Ltd.1. Introduction
Exotic materials, including copper, silver, obsidian, mica, and
shells, and mound building, both earthworks and burial mounds,
are hallmark of the Hopewell horizon (~400 BCE to 400 CE;
Prufer, 1961) within the Middle Woodland period in Eastern
North America. Although a volumetrically miniscule proportion
of the artifacts, meteoritic metal represents the most exotic and
traceable of these materials. Over the past century, meteoritic
iron has figured prominently in the discussion of exotic materials
among Hopewell sites. The movement of material betweenmy.Marquardt@Colorado.edu
Biochemistry, University of
5, USA.Hopewell sites has alternatively been envisioned as a regularized
exchange system with material moving through multiple ex-
changes before reaching their destination e the Hopewell
Interaction Sphere of Struever (1964) e or through long-distance
logistic trips to sites of known resources e the “one-shot” model
of Griffin et al. (1969).
Although artifacts of irons metal were first recognized from
Hopewell sites in modern day Ohio, including from mounds near
Chillicothe (Fig. 1), in the early 19th century (Atwater, 1820), debate
concerning their meteoritic origin began in the late 19th and early
20th century (Kinnicutt, 1884; Kunz, 1890; Farrington, 1902). The
famous Tiffany and Company gemologist George Kunz first sug-
gested a connection between numerous Hopewell artifacts from
Ohio, including beads and adze blades, and the Brenham, Kansas,
pallasite meteorite based on the shape of the included olivine
crystals and the relative proportions of metal and olivine in this
stony-iron meteorite. A definitive connection between Ohio
Hopewell meteoritic iron and the Brenham, Kansas pallasite came
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e2214with the advent of instrumental neutron activation analyses of
meteoritic irons. By determining the trace element chemical
composition of both the artifacts and metal within the Brenham
meteorite, Wasson and Sedwick (1969) convincingly demonstrated
a match. The case for a single meteoritic source was furthered by
Struever and Houart (1972) in their description of the range of
exotic materials present within the Hopewell Interaction Sphere.
Alternatively, Prufer (1961) suggested that multiple meteorites
were sampled at Hopewell sites in modern-day Ohio, consistent
with the significant span of time represented in these sites, and
Seeman (1979) argued for multiple sources of meteoritic material
within the Hopewell Interaction Sphere based on the widespread
geographic distribution of meteoritic metal artifacts. Carr and Sears
(1985) provided a more detailed assessment of Middle Woodland
meteoritic artifacts, particularly from the Southeastern United
States. These authors provided evidence for multiple sources of
meteoritic materials, noting that artifacts from Tunacunnhee
Mound E and Mandeville Mound B, both from Georgia, required
different source meteorites. More broadly, Carr and Sears (1985)
noted that numerous iron meteorite falls were likely available to
and used by the Hopewell, making it difficult to source any specific
meteorite without determination of major, minor and trace ele-
ments of the artifact in question. Halsey (1996) provides the most
comprehensive recent inventory of meteoritic artifacts from
Hopewell sites in his work on the use of metal in the Eastern
Woodlands.
A significant impediment to more fully understanding the
importance of meteorites in Hopewell culture is the absence of a
confirmed relationship between any other Hopewell meteoritic
artifact and a known meteorite, other than the Ohio Hopewell-
Brenham link. A second confirmed link might provide insights
into the mechanisms by which these exotic objects moved from the
find site to the ultimate place of archaeological deposition. In this
study, we build on preliminary work of Marquardt et al. (2008) and
McCoy et al. (2008) to present evidence supporting a link between
meteoritic metal beads from the Havana, Illinois, site and a known
meteorite find (Anoka) near modern-day Minneapolis, Minnesota,
and discuss the implications of such a link.Fig. 1. Map of find site of the Anoka iron meteorite in Anoka and Champlin, Minnesota
and the recovery site of the Havana meteoritic metal bead in Havana, Illinois. These
sites are connected via the Mississippi and Illinois Rivers. Also shown are Hopewell
centers near modern-day Chillicothe, Ohio and Trempeleau, Wisconsin. Modern state
boundaries shown for reference.2. Havana Hopewell meteoritic beads
In the summer of 1945, the Illinois State Museum excavated a
group of mounds scheduled to be leveled for the erection of a po-
wer plant (McGregor, 1952). Located in Mason County, Illinois,
about 1.5 miles south of the town of Havana (Fig. 1), the mounds
were located on the alluvial flood plain of the Illinois River. Mound
9was the largest of these structures with its base ~25 feet above the
level of the river and rose another ~30 feet to the peak. During
excavation of Mound 9, remains of 14 humans were identified. One
of these e burial 10 e was a bundle containing bones which
exhibited copper stains but no copper ornaments or implements.
Also found within this bundle were over 1000 shell and pearl beads
and a string containing twenty-two meteoritic [note: many of the
early publications used the now disfavored term “meteoric”] iron
beads with intervening shell beads (McGregor, 1952). Arnold and
Libby (1951) reported an age of 2336 ± 250 years for wood from
Havana mound #9, consistent with ages for the Ohio Hopewell
sites.
Grogan (1948) was the first study of the meteoritic beads from
the Havana site. Despite extensive weathering that had both coated
and infilled the beads with limonite, he recognized the distinctive,
yet deformedWidmanstatten pattern within the beads (Fig. 2). The
Widmanstatten pattern is the geometric pattern observed on iron
meteorites that have been cut, polished and etched. This pattern is
the result of cooling and mineral exsolution of the iron-nickel
minerals kamacite and taenite when the iron meteorite formed in
the core of an ancient asteroid. The texture is a reflection of the
chemistry and cooling history of the meteorite and has been used
as a classification tool. Grogan (1948) surmised that the beads were
produced by rolling a sheet of meteoritic metal itself produced by a
combination of cold working and mild heat treatment.
Wasson and Schaudy (1971) provided the first chemical analysis
of the Havana beads. These authors reported elemental concen-
trations of 11.4 wt % nickel, 20.5 ppm gallium, 21.6 ppm germanium
and 0.3 ppm iridium, a region of iron meteorite compositional
space that is relatively poorly populated. Based on their chemistry
and textures, the Havana meteorite beads along with three North
American iron meteorites e Anoka, Minnesota; Edmonton, Ken-
tucky; and Carlton, Texas e and the Mungindi iron from Australia
formed the new IIIC iron meteorite group, which were subse-
quently merged with group IIID to form iron meteorite group IIICD.Fig. 2. Photograph of two Havana meteoritic metal beads with a 1 cm cube for scale.
The bead on the left (7.8 g mass) is cut perpendicular to the central hole, illustrating
the extensive alteration of the bead and infilling of the central hole. The bead on the
right (4.6 g mass) is cut parallel to the central hole and exhibits a concentrically
deformed structure.
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e22 15Wasson and Kallemeyn (2002) reevaluated the classification of the
IAB and IIICD irons; most members of the former group IIIC were
placed in the IAB-sLM subgroup of what is now called the IAB
complex. Kimberlin and Wasson (1976) noted that the composi-
tions of the five irons of the IIICD meteorite group differed
distinctly from that of the Ohio Hopewell meteoritic artifacts,
supporting the hypothesis of Prufer (1961) for multiple sources of
meteoritic materials.
Kimberlin and Wasson (1976) also argued that the composition
of the Havana meteoritic beads was not a close match chemically to
any of the 300 known North American irons, suggesting that the
source was either exhausted by the Hopewell or has not been
rediscovered. As we discuss below, the most likely source meteor-
ites for the Havana beads among known meteorites exhibited no
evidence of sampling from the single collected masses. It is
important to note that Kimberlin and Wasson (1976) based their
conclusion on analyses by Wasson and Schaudy (1971), who re-
ported a limited number of elements (Ni, Ir, Ga, Ge). Further, the
influence of weathering on iron meteorite compositions was only
poorly understood at that time, as was the range of variation ex-
pected within replicate analyses of the same meteorite. Buchwald
(1976) provided a detailed metallographic description of the Ha-
vana beads, noting significant metallographic effects of combined
cold working by hammering and intermittent heat treatments to
temperatures estimated at ~650 
C, about the temperature gener-
ated by a wood-fueled fire. Buchwald (1976) and Carr and Sears
(1985) both commented on the similarity between the Havana
beads and the Carlton, Edmonton (Kentucky) and Anoka iron me-
teorites. Carr and Sears (1985) further argued that the large me-
teoritic finds Kenton County (IIIAB iron) and Mt. Vernon (pallasite),
both in Kentucky, might have served as sources for Hopewell arti-
facts. However, Kenton County differs dramatically in composition
from any known Hopewell artifact and Mt. Vernon is chemically
dissimilar to the Havana beads and was reported to have fallen in
1868.
3. Re-examining a link between Anoka and Havana
As noted by Wasson and Schaudy (1971), the Havana beads are
chemically grouped with three North American iron meteorites e
Anoka, Minnesota; Edmonton, Kentucky; and Carlton, Texas. The
first two of these were found within the broad Hopewell Interac-
tion Sphere, while Carlton, Texas e ~50 miles southwest of Dallas-
Ft. Worth e lies outside this range, but within distances traveled to
obtain exotic materials used by the Hopewell culture, such as the
distance between the Ohio Hopewell center and the Brenham,
Kansas, meteorite find site. While the Havana beads exhibited
chemical similarities with these three irons, a significant argument
against these meteorites representing the raw material for the
beads was that each was found buried as a single mass and none
showed any evidence of having pieces removed. Anoka, Minnesota,
in particular, was found during excavations about four feet beneath
the surface. In 1997, however, an additional mass of the Anoka iron
was offered for sale. Discovered in Champlin, Minnesota in 1983
while excavating a sewer, the newmass was chemically identical to
Anoka. Interestingly, Champlin and Anoka are on opposite sides of
the Mississippi River (Fig. 1) at its intersection with the Rum River.
Thus, at least two, and quite possibly, more iron masses were
deposited along a major waterway travel route.
This conclusion is supported by the projected size of the Anoka
mas, as determined through measurement of cosmic-ray produced
rare-gas isotopes (Chang and W€anke, 1969), which are produced
when a meteorite is exposed to high-energy cosmic rays in space.
Larger iron meteorites have their interiors shielded from cosmic-rays by the dense outer layers and all meteorites experience
decay of these cosmic-ray produced rare-gas isotopes after they fall
to Earth and are shielded by the Earth's atmosphere. Despite a
considerable uncertainty in their measurement by decay counting,
the unusually low 10Be of 1.6 ± 1.0 dpm/kg suggests a pre-
atmospheric radius in excess of 40 cm and associated pre-
atmospheric mass in excess of 4000 kg, far in excess of the recov-
ered mass. Based on the measured 36Cl/10Be ratio of ~5, Chang and
W€anke (1969) suggested a relatively young terrestrial age of less
than 100 ka. This limit would permit the fall of Anoka during the
most recent glaciation 15e110 ka ago. Amore direct explanation for
their occurrence is a shower of irons e meteorites that fell sepa-
rately after breakup of a larger object during atmospheric entry e
occurred in this area in the post-glacial period, producing a strewn
field of multiple samples. Unfortunately, even newer techniques for
terrestrial age dating of iron meteorites do not have the necessary
precision to distinguish between a fall during the glaciation or
during the present interglacial period.
The Mississippi River cuts through the Anoka strewn field with
erosion creating banks some 15e20 feet high. The new evidence for
recovery of multiple masses of Anoka, coupled with a re-
examination of the metallography and composition of the Havana
beads and the Anoka, Carlton and Edmonton (Kentucky) irons and
likely trade routes between these sites and Havana, Illinois, sug-
gests an Anoka mass as the parent for the Havana beads. It seems
plausible that amass of this ironwas recovered and used during the
Middle Woodland period to produce the beads found in Havana,
Illinois. To test this hypothesis, we compare the structure and
chemistry of the Havana beads and the Anoka iron and examine
evidence that suggests that Carlton and Edmonton (Kentucky) were
not parental to the beads.
4. Methods and materials
We examined three of the Havana meteoritic iron beads, two
from the collections of the Smithsonian Institution (USNM 1466)
and one from UCLA (109). Each of the beads from the Smithsonian
collection had cut and polished surfaces that were perpendicular
to the central, limonite-filled hole of the bead, with one also
having been cut and polished parallel to, but offset from, the
central hole. The bead from UCLA was examined as a polished
section and had been cut parallel to, but offset from, the central
hole, exposing a surface area of ~9  10 mm. We also examined a
polished section of Anoka (USNM 6930) from the sample exca-
vated in 1983. Both the Havana beads and Anoka iron were
examined for metallography in reflected light using a Nikon
Optiphot polarizing microscope. The major element chemical
compositions of the Anoka iron and Havana beads were deter-
mined using the JEOL 8900R electron probe microanalyzer at the
Smithsonian Institution. Four traverses e three of which were
each 3 mm in length and one of which was ~0.5 mm in length e
were conducted on both a Havana bead and the Anoka iron to
determine average iron, nickel, phosphorus and cobalt concen-
trations. Well-characterized pure element metals (Co), alloys
(Fe90Ni10) and minerals (schreibersite, troilite) were used as
standards. Microprobe totals of ~100 wt % suggest minimal alter-
ation and hydration of the material analyzed.
The polished thin section (UCLA 109) used for metallography
and major element composition of the Havana bead was analyzed
by laser ablation inductively coupled plasma mass spectrometry
(LA-ICP-MS), following the same traverses measured in the elec-
tron microprobe for major and minor element composition. Trace
element abundances were determined by laser ablation induc-
tively coupled plasma mass spectrometry (LA-ICP-MS) using
Fig. 3. Reflected light photomicrograph of Havana meteoritic metal bead exhibiting
deformed lamellae of kamacite (gray) and taenite (white with black interior) which are
concentric to the center of the bead. Black patches are areas where hydrated iron
oxides (limonite) of terrestrial origin have formed. Note that weathering preferentially
alters the low-Ni kamacite phase.
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e2216methods described in previous publications (e.g. Brenan et al.,
2003). Samples were ablated using a frequency quintupled Nd-
YAG New Wave UP213 ultraviolet laser. Ablated material was
entrained in a stream of He (ca.0.7l min1) then mixed with Ar
(ca.0.8l min1) and introduced into the plasma of a Thermo-
Finnigan Element 2 inductively coupled plasma mass spectrom-
eter (ICP-MS). Samples were analyzed by carving ablation tracks
along the line of the electron probe analyses, at a rate ranging
from 10 to 30 mm s1. Spot sizes for analyses were determined by
element concentration and sample size, with most sample ana-
lyses using an 80 mm spot. The laser fluence was controlled to
between 2e3 Jcm2 for efficient, non-thermal ablation. Samples,
reference materials (iron meteorites Coahuila, Hoba and Filomena)
and standards (metals NIST1263a, NIST1158) were analyzed in the
same fashion. Nickel, determined by electron probe analysis, was
used as an internal standard and concentrations were calculated
using modified LAMTRACE software. Isotopes were chosen to
avoid elemental and molecular interferences and tuning was car-
ried out to maximize the 1 þ ion whilst minimizing oxide
production.
In addition, separate samples of a Havana bead and the Anoka
iron were studied for bulk chemical composition at UCLA to
determine up to 16 elements (15 plus Fe) in metal by instrumental
neutron-activation analysis (INAA) in replicate analyses; data for Fe
are used for internal normalization (Wasson and Huber, 2006;
Wasson et al., 2007; Wasson and Choe, 2009). With INAA, Cr, Fe,
Co, Ni, Cu, Ga, As, Ir and Au can be measured in all iron meteorites.
As discussed in Wasson and Choe (2009), we achieve very high
precision for Co, Fe, Ga and Au (95% relative confidence limits on
means (of duplicate analyses) of 1.5e3%) and high precision (4e6%
confidence limits) on Ni, As and Ir. The analytical precision for Cr is
high (~5%) but sampling errors can be large and there is a minor
interference from the 54Fe(n,a)51Cr reaction that adds about 5 mg/g
apparent Cr to our values (which are not corrected). We achieve
confidence limits of 7e10% for concentrations higher than the
following values (in mg/g): Ge, 150; Ru, 5, Sb, 0.20, W, 0.3, Re, 0.10,
Os,1.0 and Pt, 2. A potential problemwith some of our earlier data is
W contamination resulting from cutting with tungsten-carbide
tools; in cases with large amounts of contamination, the W
values have been eliminated, but minor amounts of contamination
are difficult to spot.
To evaluate the thermal and deformation history of the Havana
beads, we performed a series of controlled heating experiments on
pieces of the Anoka iron that were cut into rectangular shapes
(~0.5  1  1 cm) using an abrasive saw. Pieces were deformed
without heating, heated at temperatures of 500e700 
C for 3 h
without deformation, and heated at 500e700 
C for 3 h with
deformation. Heating was conducted in a box furnace and defor-
mation by hammering with steel tools on a steel bench vise.
Hammering was done perpendicular to the ~1  1 cm surface until
the piece exhibited significant flattening. Samples were prepared as
polished sections and examined for metallography in reflected light
using a Nikon optiphot polarizing microscope.
We also undertook a bead making experiment using a wood-
fueled fire for heating and lithics for deformation, simulating
contemporary Hopewell technology. A cut piece of Anoka
measuring ~0.7  1  1.3 cm, containing a prominent 1 mm
schreibersite ((Fe,Ni)3P) grain and bounded on one side by the
weathered surface of the meteorite, was first subjected to defor-
mation without heating. Subsequent thinning was achieved by
three cycles of heating for ~15 min in a red-orangewood-fueled fire
interspersed with periods of deformation with lithics. This pro-
duced a final sheet 2e3 mm thick. A final round of heating was
followed by curving of the piece into the groove of a glacially-
striated lithic using another stone as a hammer. Once initiallycurved, the piece was grasped between two green twigs and easily
hammered into a cylinder around one twig. A cross section of the
produced bead was examined metallographically.5. Metallography
Our observations of the metallography of the Havana beads and
the Anoka iron confirm those of Buchwald (1976). Anoka exhibits a
fine-scale Widmanstatten structure, formed during cooling from
~600 to 300 
C when low-Ni kamacite exsolves from high-Ni
taenite. The average kamacite lamellae width in Anoka is
0.34 ± 0.05 mm. Kamacite (also called a phase) exhibits subgrain
boundaries and shock-produced, plate-shaped twin lamella called
Neumann bands decorated by micrometer-sized carbides and
taenite (also called g phase). Kamacite also exhibits a pronounced
depletion of Ni adjacent to taenite. Zoned taenite exhibits M-sha-
ped Ni profiles indicative of slow cooling, with rims reaching
~35 wt% Ni and the center of taenite lamellae exhibiting the fine
grained a þ g structure termed duplex plessite (Fig. 3). The avail-
ability of new slices of Anoka from the 1983 find reveals features
not noted by Buchwald (1976). We examined two slices with a total
surface area of ~400 cm2. We observed rare ~5 mm troilite (FeS)
nodules, whereas troilite was absent in the samples studied by
Buchwald (1976). Notably, Buchwald (1976) reported schreibersite
((Fe,Ni)3P) as skeletal crystals typically 2  0.1 mm in size. In
contrast, the larger slices of the 1983 Anoka mass include skeletal
schreibersites that typically reach 2e3 cm in length. On one slice, a
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e22 17single, elongate, skeletal schreibersite of 12 cm in length bisects the
entire slice and is truncated at the edges of the slice.
In the Havana beads, alternating bands of kamacite and taenite
are heavily deformed and roughly concentric to the center of the
bead (Fig. 3). Kamacite lamellae range up to 0.25 mm inwidth, with
intervening taenite typically 30e50 mm in width, but ranging up to
~200 mm. The kamacite is a two-phase mixture of recrystallized a
phase of ~20 mm in grain size with interspersed 0.5e2 mm gamma-
particles. This unequilibrated two-phase mixture suggests heating
during the manufacture of the beads to temperatures of
600e700 
C (e.g., the Dalton iron meteorite; Buchwald, 1976) with
formation of taenite (g) phase in kamacite along shock-produced
Neumann bands. Neumann bands in kamacite decorated by
micron-sized taenite and the carbide cohenite ((Fe,Ni)3C) are
common in Anoka, while Neumann bands in Carlton and Edmonton
(Kentucky) are rare and decorated by micron-sized phosphides
called rhabditis. The absence of cohenite in the Havana bead likely
reflects the thermal instability of the carbide. Taenite has also been
substantially altered. The taenite is zoned with respect to Ni with
M-shaped zoning profiles, but with original plessite (two-phase)
interiors homogenized (Fig. 4). Buchwald (1976) attributed the
recrystallized nature of the kamacite and absence of duplex plessite
to cold-working and reheating. Rare schreibersite are heavily
fractured and brecciated. As noted originally by Grogan (1948), the
outer surface and inner hole are filled with limonite. We note that
within the relatively unaltered middle zone, kamacite is preferen-
tially weathered relative to taenite, an observation also made by
Buchwald (1976), although this is most prominent in the section
studied along the axis of the inner hole (Fig. 3). Alteration of
kamacite is minor away from this axis and it was these portions
that were sampled for bulk chemistry by electron microprobe and
LA-ICP-MS.
6. Bulk chemistry
The Havana bead and Anoka iron yield remarkably similar major
and minor element compositions. The average for the Havana bead
is 86.9 wt % iron, 11.4 wt % nickel, 0.12 wt % phosphorus and 0.64 wt
% cobalt. The Anoka iron yielded a composition of 86.8 wt % iron,
11.3 wt % nickel, 0.04 wt % phosphorus and 0.64 wt % cobalt. In
contrast, both Edmonton (Kentucky) (12.9 wt % Ni) and Carlton
(13.2 wt % Ni) (Wasson and Kallemeyn, 2002) are richer in Ni than
either the Havana beads or the Anoka iron. This and other chemical
differences, as well as the nature of mineral phases included inFig. 4. Taenite zoning profiles in Havana and Anoka. Both exhibit M-shaped zoning
profiles. The center of Anoka taenite lamellae is duplex plessite, while the center of
taenite in Havana has been homogenized by cold-working and reheating (Buchwald,
1976).kamacite (gamma Fe, Ni metal vs. schreibersite), suggest that
Edmonton (Kentucky) and Carlton, although in the same group as
the Havana beads and Anoka iron, are not parental to the Havana
beads. The difference in phosphorus concentration between the
Havana beads and the Anoka iron is undoubtedly due to the het-
erogeneous distribution of the phosphorus-bearing mineral
schreibersite.
Compositions for the Havana bead and the Anoka iron deter-
mined by INAA are listed in Table 1. In Fig. 5 we plot some INAA
concentration data for several key elements. Shown on the dia-
gram are the members of the IAB main group (IAB-MG) and the
three low-Au subgroups IAB-sLH, IAB-sLM, and IAB-sLL. These
groups are part of the IAB complex which consists of irons that did
not experience efficient fractional crystallization. Also plotted as
small black dots are iron meteorites that have high contents of As,
Au and Sb and thus appear to be part of the IAB complex but that
differ in detailed properties from those in the groups and
subgroups.
Two of the elements, Ga and Co, show relatively small ranges
within groups but moderately large ranges among all irons. They
are thus well suited to show genetic relationships among irons. The
element Ni is the secondmost abundant element in ironmeteorites
(after Fe). Ni and Au behave as incompatibles; during crystallization
they tend to remain in the melt. On X-Y diagrams used to examine
interrelationships among iron meteorites, one commonly chooses
Au as the independent variable. Earlier, before Au data were widely
available, the data were plotted against Ni.
The element Ir is compatible; during crystallization of a melt it
always prefers to enter the solid (crystallizing) metal. Compared to
other elements, Ir tends to show large ranges within the irons
belonging to a group, and is thus well suited to distinguish these
irons from each other. Among the five plotted elements it shows the
largest range within the sLM subgroup.
Anoka (Ano) and Havana (Hav) are members of the sLM sub-
group shown by squares; to highlight them we have added circles
around the squares. The two Havana INAA replicates divergedmuch
more than is common for samples that consist of solid metal; we
attribute this to the weathering of the beads. As discussed above,
appreciable weathering of the metal (and especially the kamacite)
can be seen by microscopic examination. The first analysis has a
composition closer to that of other sLM irons, and we infer that it is
closer to the original composition prior to weathering. Both ana-
lyses are enriched in Au which we attribute to contamination of the
small samples in the museum or earlier. The data plotted in Fig. 5
are the data for the first analysis of Havana. If we assume that the
Au value of Havana is slightly high (by about 5e10%) and focus on
the other 4 elements, the agreement between Anoka and Havana isTable 1
Mean compositions of Anoka and Havana determined by INAA. The Anoka mean is
based on four analyses, weighted in the mean for the quality of the analysis. The
Havana mean is the composition determined for the first replicate; the second
replicate was similar, but showed appreciable differences attributed to weathering.
Anoka Havana
Cr mg/g 21 21
Co mg/g 5.52 5.38
Ni mg/g 119.1 125.9
Cu mg/g 196 275
Ga mg/g 18.9 22.7
As mg/g 22.1 24.9
Sb ng/g 447 434
W mg/g 0.12 0.07
Re ng/g 23 <50
Ir mg/g 0.176 0.198
Pt mg/g 0.9 0.8
Fig. 5. INAA concentration data for several key elements. Shown on the diagram are the members of the IAB main group (IAB-MG) and the three low-Au subgroups IAB-sLH, IAB-
sLM, and IAB-sLL. Small black dots are iron meteorites that have high contents of As, Au and Sb and thus appear to be part of the IAB complex. Labelled samples are Edmonton, KY
(Edm), Anoka, MN (Ano), Havana, IL (Hav) and Carlton, TX (Car).
Table 2
Bulk composition of Anoka and Havana determined by LA-ICP-MS.
Anoka Havana
P mg/g 0.65 1.31
Cr mg/g 8.1 8.6
Fe mg/g 860.020 868.511
Co mg/g 6369 6259
Ni mg/g 121.05 113.39
Cu mg/g 203 246
Ga mg/g 18.8 19.6
Ge mg/g 22.3 22
Mo mg/g 4.2 3.8
Ru mg/g 0.49 0.56
Rh mg/g 0.22 0.23
Pd mg/g 4.90 5.20
Ag mg/g 1.5 10.9
W mg/g 0.11 4.20
Os mg/g 0.23 0.25
Ir mg/g 0.16 0.16
Pt mg/g 0.23 0.22
Au mg/g 1.5 1.7
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e2218remarkably close, the more so considering the weathered condition
of the Havana beads.
As discussed above, the other North American irons with com-
positions relatively similar to Havana are the two IAB-sLM irons
from Edmonton, KYand Carlton, TX. These are also in the set of sLM
data plotted as squares on Fig. 5. Havana is similar to both of these
but its Ir content differs by a factor of 2 ormore; that in Edmonton is
0.49 mg/g, 2.3  higher, that in Carlton is 0.082 mg/g, 2.6  lower. In
contrast, the Ir content of Havana is 1.2 higher than that in Anoka;
such variations are not uncommon when separate masses of me-
teorites from the IAB complex are analyzed.
In summary, the INAA compositional data of Havana is close
enough to that of Anoka to be fully consistent with the view that an
iron from the Anoka shower was the source of the Havana beads.
Laser ablation inductively coupled plasma mass spectrometry was
used to determine a broader array of elements, although obtaining
a representative bulk analysis is more challenging. Compositions
for the Havana beads and the Anoka iron from LA-ICP-MS are
shown in Table 2 and Fig. 6a. In general, elements determined by
both methods are in reasonable agreement, although concentra-
tions of Cr and Pt differ by factors of 3e4. As with the INAA data, the
Havana beads and the Anoka iron are remarkably similar in
chemical composition (Fig. 6b), with most elements within 10%
between Anoka and Havana. Exceptions are elements concentrated
in the phosphate schreibersite (P, Ag; possibly W (Corrigan et al.
(2009))) and elements that may have been contaminated by
other artifacts (e.g., Cu). A much smaller difference inW is observed
in the INAA data, suggesting that schreibersite concentrations differbetween the Havana beads and the Anoka iron in bulk, although to
a much smaller extent than observed in the LA-ICP-MS data.7. Manufacture of the Havana beads
The basic manufacturing of the Havana beads was described by
McGregor (1952). The beads weremade by cold-workingmetal into
Fig. 6. Composition of the Anoka iron meteorite and Havana bead determined by LA-ICP-MS, normalized to CI chondrites (A) and rationed to each other for those elements with
ratios between 0.8 and 1.2 (B). The composition of the two are remarkably similar with marked differences in elements concentrated in phosphides (P, Ag, possibly W). This data
strongly supports that the Anoka iron meteorite was the parent material from which the Havana beads were made.
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e22 19strips and joining the ends to form a hollow, tubular bead, which
was subsequently ground to produce the globular shape. While this
basic description is consistent with the morphology of the beads, it
does not address how a single, malleable mass of iron was sepa-
rated to form 22 beads or the exact process by which mixed cold-working and heating occurred. Separating smaller pieces of metal
from an iron meteorite could be accomplished with repeated cold-
working and deformation, but may have been achieved in other
ways as well. In the case of the Brenham meteorite that is parental
to the Ohio Hopewell meteoritic beads, the metal forms an
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e2220interconnected network between brittle olivine crystals. Weath-
ering or physical degradation of the olivine crystals leaves a
meshwork of metal from which small pieces are relatively easily
removed. In contrast, the Anoka iron occurs as coherent, large
masses of iron. However, Anoka also contains large crystals of
schreibersite which, in some cases, cross cut the iron (Fig. 7). The
schreibersite is substantially more brittle than the surrounding
metal, perhaps providing a point of weakness from which smaller
pieces were removed from a larger mass. Buchwald (1976)
observed the limited abundance of schreibersite in the Havana
beads, consistent with our observations, and suggested that the
beads might have been produced from pieces of metal between the
larger schreibersites.
Buchwald (1976) argued that the structures present in the
Havana beads suggested both cold-working and intermediate
annealing through heating, with the absence of typical a2
structure limiting the upper temperature of heating to
750e800 
C. Among our controlled heating and deformation
experiments, the combination of deformation with heating to
700 
C for 3 h most closely duplicated the Havana bead, althoughFig. 7. Slice of the Anoka iron exhibiting prominent schreibersite, including a
0.5  12 cm skeletal crystal that bisects the slice, and a fine Widmanstatten pattern.
Maximum dimension of the slice is 26 cm.
Fig. 8. Meteoritic metal bead formed from the Anoka iron using wood-fueled fire for heating
Scale cube is 1 cm on a side. Left: Reflected light photomicrogaph of a cross section of the be
diameter.the duplex plessite was not homogenized and the grain size
within the recrystallized kamacite was much larger in our
experiment (~200 mm) than in the Havana bead (~20 mm). These
differences likely reflect the repeated heating and deformation
cycles experienced during manufacture of the Havana bead in
contrast to the single heating and efficient deformation in our
experiment using controlled heating and deformation with steel
tools.
In our experiments using wood-fueled fire and lithics, cold-
working in the absence of heating produced extensive fragmenta-
tion of the schreibersite and modest thinning of the piece at the
margins, suggesting that removal of distinct pieces by fragmenta-
tion of schreibersite is a viable mechanism. The bead produced by
repeated cycles of heating and deformation (Fig. 8) shares a number
of similarities with the Havana bead, including curvature and
deformation of the Widmanstatten pattern, fragmentation of the
schreibersite, and, notably, subgrains in kamacite that are of
approximately the same scale (~20 mm) and are decorated by
1e2 mm gamma (g) particles. Duplex plessite has not been ho-
mogenized in the produced bead, perhaps suggesting annealing for
longer periods of time or at slightly higher temperature for the
Havana beads.
8. Anoka to Havana
The movement from source to final location for these meteoritic
beads provides an interesting test of the competing models of the
Hopewell Interaction Sphere and the “one shot” expedition model.
While most of the exotic materials (e.g., obsidian, mica, copper)
used by the Hopewell occurred in abundance at their sources, that
is not universally true of meteorites. The Brenhammeteorite, which
is the source of the meteoritic metal beads identified at the
Hopewell mounds in Ohio, has produced many tons of material up
to the present. As such, the idea that an expedition might visit that
site for the specific purpose of returning material to Ohio seems
tenable.
In contrast, the 22 metal beads identified from the Havana
mound almost certainly originated from a single mass and addi-
tional masses were likely not known to the Hopewell. The
pathway from the fall location of the Anoka meteorite in Minne-
sota and the final location in Havana seems straightforward. The
Anoka meteorite fell near the Mississippi and the Havana mounds
were located along the banks of the Illinois River, a tributary of theand lithics for deformation. Right: Cylindrical bead formed by rolling a flattened sheet.
ad illustrating the deformation of the Widmanstatten pattern. Bead is 1 cm in exterior
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e22 21Mississippi (Fig. 1). In this case, the idea of exchange seems more
likely, although the exchange of raw material may have occurred
between neighboring groups rather than through the series of
complex, regularized exchange envisioned as the Hopewell
Interaction Sphere. The Havana Hopewell center likely interacted
with both the adjacent Trempeleau Hopewell (Fig. 1), which
extended from southwestern Wisconsin up the Mississippi river
to the find site of the Anoka meteorite, as well as the more distant
Ohio Hopewell which centers in southern Ohio. Given the singular
nature of the Anokamass used to form the Havana beads, the most
likely scenario for recovery was by local inhabitants who were
part of the Trempeleau Hopewell. Named for mounds near mod-
ern Trempeleau, Wisconsin, this expression of the Hopewell
constructed mounds now found in St. Paul, Minnesota, ~50 km
down the Mississippi River from Anoka. The St. Paul mounds were
investigated by T.H. Lewis in 1879 (Lewis, 1896a,b), who, among
other artifacts, reported the presence of two pieces of hammered
native copper. The connection of the Trempeleau and Havana
Hopewell via the Mississippi further supports recovery of the
Anoka mass by the Hopewell and, ultimately, trade to the Havana
center.
Where the beads were manufactured is poorly constrained.
Metal working was common among the Hopewell, with copper
being the most commonly worked metal to manufacture a variety
of ceremonial objects including beads. Copper beads are known
from Ohio, Havana and Trempeleau sites (Kocik, 2012), likely
reflecting the exchange of stylistic concepts as envisioned as part of
the Hopewell Interaction Sphere (Struever, 1964). Among these
sites, the most diverse examples of copper working, coupled with
the only other known example of meteoritic metal beads, might
suggest that the beads were manufactured by the Ohio Hopewell.
This would be consistent with the general paucity of copper objects
in the Havana mound where the meteoritic metal beads were
excavated (McGregor, 1952), suggesting copper working expertise
in the Havana Hopewell center. However, given that copper beads
have been identified in all three Hopewell centers, the direct
connection between Anoka and the Havana site via the Mississippi
and Illinois rivers, and the absence of comparable composition
meteoritic beads among either the Trempeleau or Ohio Hopewell
sites (Fig. 1), the most likely scenario suggests manufacture by the
Havana Hopewell.
9. Summary
Spurred by the recovery of additional masses of the Anoka
iron and the realization that a shower of irons occurred near the
Mississippi River, we have examined whether the Hopewell
meteoritic beads recovered from Havana, Illinois can be uniquely
linked to the Anoka fall. The similarity in major, minor and trace
element chemistry between Anoka and Havana, the presence of
micrometer-sized inclusions of gamma iron in kamacite in both,
and the obvious connection via the Mississippi and Illinois Rivers
between Anoka and Havana point to the likely production of the
Havana beads from a mass of the Anoka iron. In contrast, Carlton
and Edmonton (Kentucky) differ in major element chemistry,
particularly Ni, and in the nature of mineral phases included in
kamacite (gamma Fe, Ni metal vs. schreibersite) from the Havana
beads. Our experiments strongly support the manufacture of the
beads via fragmentation of schreibersite to liberate small pieces
of metal followed by repeated cycles of heating to temperatures
of 600e700 
C followed by cold-working to produce a flattened
metal sheet which was subsequently rolled to make the Havana
beads. Recovery of the individual mass of Anoka used to make
the beads was probably by persons associated with the Trem-
peleau Hopewell center, with exchanges bringing this material tothe Havana Hopewell center, where the beads were
manufactured.
Conflicts of interest
The authors declare that they have no conflict of interest.
Acknowledgments
The authors thank NASA's Cosmochemistry and Emerging
Worlds Programs for financial support of this work, the National
Science Foundation REU Program for support for AEM during her
undergraduate internship at NMNH/SI, Catherine M. Corrigan,
Nicole Lunning, Bruce D. Smith and R. Eric Hollinger (NMNH/SI) for
consultation during the study and review of the manuscript, and
two anonymous reviewers whose comments helped clarify the
manuscript. Kees Welten provided particularly useful insights into
the history of the Anoka iron based on rare gas analyses. We also
thank Prof. William F. McDonough in the Dept. of Geology at the
University of Maryland for advice, support, and access to the ICP-
MS laboratory.
References
Arnold, J.R., Libby, W.F., 1951. Radiocarbon dates. Science 113, 111e120.
Atwater, C., 1820. Description of the antiquities discovering in the State of Ohio.
Trans. Collect. Am. Antiq. Soc. Archaeol. Am. 1, 168e178.
Brenan, J.M., McDonough, W.F., Dalpe, C., 2003. Experimental constraints on the
partitioning of rhenium and some platinum-group elements between olivine
and silicate melt. Earth Planet. Sci. Lett. 212, 135e150.
Buchwald, V.F., 1976. Handbook of Iron Meteorites. University of California Press,
Berkeley.
Carr, C., Sears, D.W.G., 1985. Toward an analysis of the exchange of meteoritic iron in
the Middle Woodland. Southeast. Archaeol. 4, 79e92.
Chang, C., W€anke, H., 1969. Berryllium-10 in iron meteorites, their cosmic-ray
exposure and terrestrial ages. In: Millman, P.M. (Ed.), Meteorite Research.
Springer-Verlag, New York, NY, pp. 397e406.
Corrigan, C.M., Chabot, N.L., McCoy, T.J., McDonough, W.F., Watson, H.C.,
Saslow, S.A., Ash, R.D., 2009. The iron-nickel-phosphorus system: effects on the
distribution of trace elements during the evolution of iron meteorites. Geochim.
Cosmochim. Acta 73, 2674e2691.
Farrington, O.C., 1902. Meteorite Studies e I. Field Columbian Museum Geological
Series Pub, vol. 64, pp. 310e314.
Griffin, J.B., Gordus, A.A., Wright, G.A., 1969. Identification of the sources of Hope-
wellian obsidian in the middle west. Am. Antiq. 34, 1e14.
Grogan, R.M., 1948. Beads of meteoric iron from an Indian mound near Havana,
Illinois. Am. Antiq. 13, 302e305.
Halsey, J.R., 1996. Without forge or crucible: aboriginal Native American use of
metal and metallic ores in the Eastern woodlands. Mich. Archaeol. 41, 1e23.
Kimberlin, J., Wasson, J.T., 1976. Comparison of iron meteoritic material from Ohio
and Illinois Hopewellian burial mounds. Am. Antiq. 41, 489e493.
Kinnicutt, L.P., 1884. Report on the meteoric iron from the altar mounds in the Little
Miami Valley, Ohio. Rep. Peabody Mus. Am. Archaeol. Ethnol., Camb. 3,
381e384.
Kocik, C., 2012. Regional variation in Hopewell copper use. Univ. Wisconsin-
LaCrosse J. Undergrad. Res. 15, 1e29.
Kunz, G.F., 1890. On five new American meteorites. Am. J. Sci. 40, 312e323.
Lewis, T.H., 1896a. Pre-historic remains at St. Paul, Minnesota. Am. Antiq. Orient. J.
18, 207e210.
Lewis, T.H., 1896b. Mounds and stone cists at St. Paul, Minnesota. Am. Antiq. Orient.
J. 18, 314e320.
Marquardt, A., McCoy, T., Vicenzi, E., Ash, R., 2008. The source of Hopewell extra-
terrestrial metal and its anthropological implications. In: Abstracts of the SAA
73rd Annual Meeting, vol. 359.
McCoy, T.J., Marquardt, A.E., Vicenzi, E.P., Ash, R.D., Wasson, J.T., 2008. Meteoritic
metal beads from the Havana, Illinois, Hopewell Mounds: a source in Minnesota
and implications for trade and manufacture. In: Lunar and Planetary Science
Conference, vol. 39. #1984.
McGregor, J.C., 1952. The Havana site. In: Deuel, T. (Ed.), Hopewellian Communities
in Illinois, pp. 43e92. Illinois State Museum, Scientific Papers 5.
Prufer, O.H., 1961. Prehistoric Hopewell meteorite collecting: context and implica-
tions. Ohio J. Sci. 61, 341e352.
Seeman, M.F., 1979. The Hopewell Interaction Sphere: the Evidence for Inter-
regional Trade and Structural Complexity. Indiana Historical Society. Prehis-
toric Research Series 5(2).
Struever, S., 1964. The Hopewell interaction Sphere in riverine-western great lakes
cultural history. In: Caldwell, J., Hall, R. (Eds.), Hopewellian Studies, pp. 85e106.
T.J. McCoy et al. / Journal of Archaeological Science 81 (2017) 13e2222Illinois State Museum, Scientific Papers 12.
Struever, S., Houart, G.L., 1972. An analysis of the Hopewell interaction sphere. In:
Wilmsen, E. (Ed.), Social Exchange and Interaction. University of Michigan,
pp. 47e80. Museum of Anthropology, Anthropological Paper 46.
Wasson, J.T., Sedwick, S.P., 1969. Possible sources of meteoritic material from
Hopewell Indian burial mounds. Nature 222, 22e24.
Wasson, J.T., Schaudy, R., 1971. The chemical classification of iron meteorites e V.
Groups IIIC and IIID and other irons with germanium concentrations between 1
and 25 ppm. Icarus 14, 59e70.
Wasson, J.T., Kallemeyn, G.W., 2002. The IAB iron-meteorite complex: a group, fivesubgroups, numerous grouplets, closely related, mainly formed by crystal
segregation in rapidly cooling melts. Geochim. Cosmochim. Acta 66,
2445e2473.
Wasson, J.T., Huber, H., 2006. Compositional trends among IID irons; their possible
formation from the P-rich lower magma in a two-layer core. Geochim. Cos-
mochim. Acta 70, 6153e6167.
Wasson, J.T., Choe, W.H., 2009. The IIG iron meteorites: probable formation in the
IIAB core. Geochim. Cosmochim. Acta 73, 4879e4890.
Wasson, J.T., Huber, H., Malvin, D.J., 2007. Formation of IIAB iron meteorites. Geo-
chim. Cosmochim. Acta 71, 760e781.