ARTICLE IN PRESS

Quaternary Science Reviews 24 (2005) 1797–1808

Extending the radiocarbon calibration beyond 26,000 years

before present using fossil corals
Tzu-chien Chiua,, Richard G. Fairbanksa,b, Richard A. Mortlockb, Arthur L. Bloomc
a
Department of Earth and Environmental Sciences, Columbia University, New York, NY 10027, USA
b
Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY 10964, USA
c
Department of Earth and Atmospheric Sciences, Cornell University, Ithaca, NY 14853, USA
Received 10 January 2005; accepted 1 April 2005

Abstract

Fossil coral is an excellent archive to extend the radiocarbon (14C) calibration beyond tree-ring records; however, published coral
data older than 26,000 years before present are too disparate for practical calibration. We propose an explanation for these
discrepancies: trace amounts of secondary calcite and organic matter in fossil corals shift the 14C ages toward younger ages. In a
series of acid-leaching experiments to simulate the standard cleaning procedure prior to 14C analysis, we measure up to a 300%
enrichment of diagenetic calcite in corals due to the solubility differences between calcite and aragonite. In model calculations, we
show the 14C age offsets produced by acid leaching could be hundreds to thousands of years when typical amounts of calcite are
present in samples. We demonstrate the necessity and our ability to detect o0.2% calcite in aragonite by X-ray diffraction and
apply ‘‘o0.2% calcite’’ as our a priori criterion for our coral samples. The estimated age offsets of samples with o0.2% calcite fall
within the reported analytical uncertainty of 14C dating. In addition, we present 14C results from an extended hydrogen peroxide
(H2O2) pretreatment for coral samples to remove organic matter. Our data show that treated samples yield consistent or older 14C
dates compared to non-treated samples. These rigorous but necessary screening and cleaning techniques provide precise,
reproducible and accurate radiocarbon calibration data.
r 2005 Elsevier Ltd. All rights reserved.

1. Introduction (Elsasser et al., 1956; Lal, 1988; Bard et al., 1990a;

Guyodo and Valet, 1999; Laj et al., 2000, 2002, 2004)
The importance of the radiocarbon (14C) chronometer and to a lesser degree carbon cycling (Edwards et al.,
for age determinations in the fields of geology, 1993; Hughen et al., 2000).
paleoceanography, climate studies and archeology is Conversion of reported 14C ages to calendar ages
well documented (Stuiver et al., 1986). The reported 14C requires a calibration data set and a statistical model to
ages, however, are not accurate because the initial compute calendar age and uncertainty. Tree rings have
activity of 14C in the atmosphere is not constant with been the most useful calibration archive up to
11,500
time, and thus a basic principle of radioactive dating years before present (BP), but older samples are not
cannot be applied (de Vries, 1958; Suess, 1970; Damon easily obtained (Stuiver et al., 1998). Fossil corals are
et al., 1978). The atmospheric 14C concentration, and regarded as one of the best candidates for extending the
hence a sample’s initial 14C activity, varies with solar radiocarbon calibration (Bard et al., 1990a) beyond the
activity (Stuiver, 1961; Stuiver and Quay, 1980), tree-ring records (Stuiver et al., 1998) because corals can
shielding effect from the Earth’s geomagnetic field be directly dated by both 14C and 230Th/234U/238U
methods with minimal assumptions and does not
Corresponding author. Tel: +1 845 365 8499. involve correlations to other climatic proxies (Bard
E-mail address: tcchiu@LDEO.columbia.edu (T.-C. Chiu). et al., 1990a, 1993; Burr et al., 1998; Cutler et al., 2004;

0277-3791/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.quascirev.2005.04.002
 ARTICLE IN PRESS
1798 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

Paterne et al., 2004; Fairbanks et al., 2005). In addition, calcite (Yokoyama et al., 2000). The IntCal04 workshop
measured radiometric ages of corals have computed recommends using a limit of p1% calcite for inclusion
error estimates, which is not the case for most other in the radiocarbon calibration curve (Reimer et al.,
archives used for extending radiocarbon calibration. 2002) based on typical detection limits reported in the
Coral dates are completely independent to one another literature. Edwards et al. (1997) proposed 231Pa/235U
and so are their error estimates, unlike radiocarbon dating of coral samples as a rigorous test of the accuracy
calibrations based on laminated sediments (Hughen et of the 230Th/234U/238U age. This redundant dating
al., 1998; Goslar et al., 2000; Kitagawa and van der technique has been applied to fossil coral and used for
Plicht, 2000) and interpolated ages based on assumed radiocarbon calibration in the age range of 12–50 kyr
sedimentation rates and correlations to other climatic (Cutler et al., 2004; Mortlock et al., 2005). However, the
230
proxy records measured in ice cores (Hughen et al., Th/234U/238U and 231Pa/235U concordancy test con-
2000, 2004; Voelker et al., 2000) in which correlation firms the accuracy of the calendar age but not the
uncertainty and accumulated errors may be unavoid- radiocarbon age. The observation that the majority of
able. Additionally, it is critical to keep the ice core calibration data (Yokoyama et al., 2000) (Fig. 1) show
record of atmospheric changes, especially the atmo- considerable scatter and fall toward much younger
spheric gas chemistry, independently dated from the radiocarbon ages suggests a source of contamination,
radiocarbon calibration curve in order to be able to possibly from inadequate screening criteria and/or
measure leads and lags between radiocarbon-dated cleaning prior to 14C dating. In this study, we provide
records and ice core records. Although speleothems paired 230Th/234U/238U and 14C measurements and
can also be independently dated by U-series and 14C experimental evidence from chemically etched coral
methods with high precision, the assumed initial 230Th samples which suggest that the observed scatter in the
and dead carbon fraction (Beck et al., 2001) introduce published coral data in the period of 30–50 ka likely
some additional uncertainties and these two components results from the presence of minor amounts of
may vary with time (Reimer et al., 2002). In order to diagenetic calcite and potentially organic contamination
compute the estimated calendar age error on a radio- in fossil corals. We further demonstrate that more
carbon age conversion to calendar ages, it is necessary stringent screening of diagenetic calcite (less than
that the calendar and 14C age errors for the calibration 0.2 wt%) can be achieved and is required.
curve are precisely measured (Fairbanks et al., 2005). Diagenetic calcite measured in corals is most likely
Published coral data display significant scatter beyond precipitated from percolating rainwater during periods
26,000 years BP (Yokoyama et al., 2000) (Fig. 1). These of lower sea level or due to island uplift, although the
corals were exposed to sub-aerial conditions due to exact time history may not be well known. Although
island uplift and during the glacial lowstand, and are some authors have adopted acceptable concentrations of
reported to contain between 0% and 3% diagenetic calcite in the range of 1–3% in fossil corals (Table 1),
any calcite precipitated in coral is secondary and has the
potential for producing coral 14C ages significantly
Papua New Guinea (Yokoyama et al. 2000)
New Guinea (Bard et al., 1998) younger than the ‘‘true radiocarbon age’’. We suggest
50000 Santo and Papua New Guinea (Cutler et al., 2004)
Equiline that relatively small amounts of diagenetic calcite
become a more serious radiocarbon contaminant during
45000
the routine etching step just prior to graphite target
Conventional 14C age (years BP)

preparation for accelerator mass spectrometry analysis

40000 for 14C. Standard practices in radiocarbon laboratories
call for a 50–60% acid etch of coral samples just prior to
35000 dissolution of coral samples for the graphite target
preparation (Bard et al., 1990b, 1993; Burr et al., 1992;
30000 Edwards et al., 1993; Yokoyama et al., 2000). The
pretreatment is believed necessary for removing surface
25000
contamination and absorbed modern carbon. Burr et
al.’s (1992) partial dissolution experiments were con-
ducted on two samples older than 50 ka, both of which
20000
30000 35000 40000 45000 50000
had detectable calcite in excess of 2 wt%. It was
230
Th/234U/238U age (years BP) observed that a significant fraction of the radiocarbon
was found in the initial 10% fraction dissolved and that
Fig. 1. Published radiocarbon and U-series ages for fossil corals older
background levels of radiocarbon could only be
than 30,000 years BP (Bard et al., 1998; Yokoyama et al., 2000; Cutler
et al., 2004). The equiline is presented only for reference. Samples in obtained when more than 50% of the original sample
this figure are reported to contain diagenetic calcite concentrations in had been dissolved. Yokoyama et al. (2000) determined
the range of 0–3%. that coral samples reached a 14C age ‘‘plateau’’ after
 ARTICLE IN PRESS
T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808 1799

Table 1
Examples of the calcite contamination acceptance criteria in fossil corals selected for radiocarbon dating in the age range of 26,000–50,000 years BP
(calendar year)

Sample locality Reference source Acceptance % calcite in corals

Papua New Guinea Yokoyama et al. (2000) o3

N/A Reimer et al. (2002) o1 suggested (IntCal04 workshop)
Barbados, Kiritimati, Santo and Araki Island This study and Fairbanks et al. (2005) o0.2
Vanuatu and Papua New Guinea Cutler et al. (2004) o1

50% leaching and concluded that most contamination homogenized in an agate mortar to make a series of
had been eliminated. XRD standards. XRD slides were made by mounting
Aragonite is more soluble than calcite (Chave et al., the well-mixed standard powders on microscope glass
1962; Milliman, 1974a), therefore chemical leaching covers (1 in diameter) using a mixture of one part of
could potentially have the adverse effect of enriching the DucoCement diluted with 10 parts of acetone. Slides
mass fraction of calcite in coral, which would result in made with diluted DucoCement give comparable counts
biasing radiocarbon results toward younger ages. to conventional water-mounted slides but DucoCement
Organics are another source of contamination and are slides can be stored indefinitely, allowing use of the same
particularly common in corals from outcrop samples standard slides over time to establish instrument
exposed to soils or from mold colonization during reproducibility. In order to obtain reproducible results,
sample transit and storage. We observe that typical it is necessary to produce slides of relatively uniform
outcrop samples continue to whiten after several days of thickness, grain size, and of sufficient mass to obtain
hydrogen peroxide (H2O2) treatment in an ultrasonic optimal intensity. We determined that a standard or
bath that we attribute to continued oxidation of organic sample mass of
8 mg was required to yield a peak
matter. The potential for using fossil coral to accurately height value of
2000 counts/s which we determined to
extend radiocarbon calibration beyond 26,000 years BP be the minimum intensity for the aragonite major peak.
ultimately relies on both selecting fossil corals of only The standard powder is blended into the DucoCement/
the best preservation (i.e. no detectable calcite) and acetone mixture and evenly smeared onto a 1-in glass
aggressive pretreatment methods that effectively remove cover slide (Fisherbrand Microscope Cover Glass: 12-
resistant organic carbon. 546-2 25CIR-2) to cover 90% of the central area. The
We propose a two-step procedure before 14C dating a slide dries rapidly and should take on an opaque
coral sample, a prior X-ray diffraction (XRD) screening appearance. Nine XRD standards with various calcite
and an extended hydrogen peroxide cleaning, to concentrations (weight percentage: 0%, 0.1%, 0.2%,
circumvent these problems. (1) We apply ‘‘o0.2% 0.3%, 0.4%, 0.5%, 1.0%, 1.5% and 2.0%) were
calcite’’ by XRD analysis as the screening criterion for prepared and were routinely measured with each set of
fossil corals. (2) Sub-samples of fossil corals that pass coral samples.
‘‘o0.2% calcite’’ criterion are subjected to an extended Corals were submerged in deionized water and
hydrogen peroxide pretreatment to remove any poten- sampled using a 5 mm ID diamond corer in order to
tial organic matter. Dry-down steps of samples after this avoid formation of calcite that is produced by dry
pretreatment are eliminated to avoid any mineral drilling (Gill et al., 1995). The sub-sample was placed in
precipitation. Additional leaching (50–60%) and a final deionized water and cleaned in a high power ultrasonic
hydrogen peroxide treatment are recommended imme- bath (Fisher Model 500 Sonic Probe) for 2 h. Samples
diately prior to preparation of graphite targets in 14C were dried at T ¼ 60 1C, and then were crushed and
laboratories (Nadeau et al., 2001) ground in an agate mortar to an ultra-fine powder.
Sample slides were prepared from powdered samples in
the same manner as the standards. Test shows that no
2. Methods calcite was produced from aragonite during the coring
or grinding step in the method described here.
2.1. X-ray diffraction standards and samples
2.2. X-ray diffraction instrumentation
A pure aragonite end member was prepared from a
modern coral specimen (Acropora hyacinthus) collected XRD analyses were performed using a Philips X’pert
from Kiritimati Island in 1997. A pure calcite end X-ray Diffractometer (PW3040-MPD) located at the
member was prepared from NBS-19 marble. Aragonite Lamont-Doherty Earth Observatory (LDEO). This
and calcite end members were ground in an agate instrument is equipped with a y2y goniometer and a
mortar separately, and then they were combined and computer-controlled, y-compensating, variable-beam,
 ARTICLE IN PRESS
1800 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

slit system that allows a constant sample surface to be slide may or may not be identified during the scan, and
irradiated at all 22y angles during the diffraction scan. therefore ‘‘o0.2% calcite’’ was conservatively selected
Aragonite and calcite were measured at the major peaks as our detection limit.
3.396Å[111] and 3.035Å[104], respectively. Instrument
‘‘dwell times’’ and ‘‘22y step size’’ were selected to 2.3. Radiocarbon and uranium-series dating methods
optimize sensitivity and detection limits. Instrument
settings permitting sufficient resolution of the calcite Radiocarbon dates of coral samples in this study were
peak at o0.2% calcite in aragonite require a dwell time measured either at the Center of Accelerator Mass
of 1 s and a 22y step size of 0.02. These settings produce Spectrometry (CAMS) at Lawrence Livermore National
an analysis time of
5 min for each slide and permit Laboratory, University of California or at the Leibniz
analysis of
7 samples/h. In an attempt to improve Laboratory for Dating and Isotope Analyses, Christian-
detection limits, we explored other combinations of Albrechts-Universität, Kiel, Germany. The target pre-
‘‘dwell time’’ and ‘‘22y step size’’. We found that a paration methods, including sample pretreatment steps,
combination of a dwell time of 2 s and a 22y step size of are detailed elsewhere (Fairbanks et al., 2005). All
0.01 (
20 min analysis time) produced aragonite peaks uranium-series dates of coral samples in this study were
with greater definition but no better resolution of the obtained by using a multiple-collector ICPMS (PLAS-
calcite peak. We also investigated and rejected the MA 54) at LDEO (Mortlock et al., 2005). Our entire
combination of a dwell time of 1 s and a 22y step size of coral data set and radiocarbon calibration (0–50,000
0.01 (analysis time of
10 min) because standards years BP) are presented in detail elsewhere (Fairbanks
displayed increased scatter with these settings. et al., 2005).
The relative intensity of each component during XRD
analysis is generally proportional to its amount in the
powder mixture. Relative intensities were calculated by 3. Results
peak heights in this study although peak areas can also
be used. We compared the relationship between weight 3.1. Chemical leach
percentage of calcite versus the peak height ratio of
(calcite)/(calcite+aragonite) (Milliman, 1974b) for our We test the hypothesis that standard chemical
XRD standards. The XRD standard data can best be leaching procedures used by radiocarbon labs preferen-
described by a regression line: (calcite)/(calcite+ara- tially increase the fraction of diagenetic calcite in fossil
gonite) ¼ 0.0394  (% calcite)+0.0033 (R2 ¼ 0:9907) corals. Leaching experiments were conducted on a
(Fig. 2) based on the pooled mean of our nine XRD calcite-bearing fossil coral sample (
48,000 years BP)
standards. We estimated our analytical limits of detec- determined to have a variable calcite content in the
tion at 0.11% calcite by calculating the 3s uncertainty range of 0.5–2.7%. We replicated the leach pretreatment
about our ‘‘0%’’ calcite standard. However, during the of coral samples used at CAMS at Lawrence Livermore
course of this study, the calcite peak of a 0.1% standard National Lab by preparing an acid solution consisting
of one part of 1 N HCl and three parts of 30% H2O2 to
0.10 yield a final normality of
0.25 N.
(calcite)/(calcite+aragonite) (peak height ratios)

0.09
XRD standards measurements In one experiment, samples were divided in two parts
Linear Fit of XRD standards
with one-half subjected to a predetermined volume of
0.08
the acid solution sufficient to dissolve approximately
0.07 20%, 40%, 60% and 80% of the original sample mass.
0.06 Samples were weighed and then allowed to dissolve
0.05
overnight after which they were rinsed with deionized
water three times, dried at T ¼ 60 1C, and re-weighed.
0.04
All samples, treated and untreated were subsequently
C/(C+A)= 0.0033 + 0.0394* (calcite weight %)
0.03 crushed in an agate mortar and prepared for XRD
R2 = 0.9907
0.02 analysis (Table 2). In three out of four cases, the weight
0.01 fraction of calcite increased in samples subjected to acid
leaching (a factor of 1.3–3.4) when compared to the
0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
untreated sample half.
calcite weight % in aragonite-calcite mixed standards In the second experiment, the amount of acid
necessary to dissolve 60% of the original sample mass
Fig. 2. Standards calibration curve of % calcite in aragonite by X-ray
was added. In four out of five cases, the weight fraction
diffraction. (Calcite)/(calcite+aragonite) ratios were calculated by
peak height. Error bars represent the 1s standard deviation about the of calcite increased by a factor of 1.6–3.1 in the acid-
mean value (minimum of eight replicates). Detection limit is smaller treated samples (Table 2). Our experimental results
than 0.2% calcite (see text for details). suggest that pretreatment steps designed to remove
 ARTICLE IN PRESS
T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808 1801

Table 2 25000
Enrichment of secondary calcite in Araki coral AK-D-2 during acid
leaching 24900

Apparent 14C age (years BP)

24800
Sub-sample % Calcite % Weight % Calcite Calcite
# (untreated) loss (treated) enrichment 24700
factor
24600
AK-D-2 #1 0.9 19 2.4 2.7
AK-D-2 #2 1.3 38 1.7 1.3 24500
AK-D-2 #3 0.5 54 1.6 3.4
24400
AK-D-2 #4 1.4 70 0.5 0.3
AK-D-2 #5 2.0 54 3.2 1.6 24300 0.2 % initial calcite
AK-D-2 #6 0.5 56 1.0 2.0 0.5 % initial calcite
1.0 % initial calcite
AK-D-2 #7 0.6 56 1.3 2.3 24200 2.0 % initial calcite
AK-D-2 #8 0.7 56 2.3 3.1 3.0 % initial calcite

AK-D-2 #9 2.7 56 1.4 0.5 24100

0.0 0.1 0.2 0.3 0.4 0.5 0.6
Note: Sub-samples (#1–#9) represent replicate drilled cores (5 mm ID (a) Fractional weight loss
diamond corer operated with the slab submerged in deionized water).
Treated samples were leached with a mixed acid of
0.25 N HCl/ 45000

22% H2O2.
44000

43000
contaminant radiocarbon are likely to enrich the mass Apparent 14C age (years BP)
fraction of secondary carbonate. We speculate that the 42000
cause of variable enrichments may be due to hetero-
geneous calcite contamination observed microscopically 41000
and variations in the skeletal microstructure.
40000
14
3.2. Modeling C age offsets 39000 0.2 % initial calcite
0.5 % initial calcite
1.0 % initial calcite
In an attempt to quantify the effect of secondary 38000 2.0 % initial calcite
3.0 % initial calcite
calcite on the reported 14C age of fossil corals and the 37000
sensitivity of calculated 14C ages to a hypothetical 0.0 0.1 0.2 0.3 0.4 0.5 0.6
chemical pretreatment, we applied a calcite addition (b) Fractional weight loss
model and calculated the apparent 14C age for two
Fig. 3. (a) Apparent 14C age as a function of fractional weight loss and
scenarios. In the first scenario, we assumed a true 14C initial calcite percentage in a hypothetical fossil coral. The ‘‘true’’ 14C
age of 25,000 years BP and that calcite was added from age in each example is assumed to be 25,000 years BP. We assume the
25,000 to 12,500 years BP (14C year) (Fig. 3a). The calcite is added from 25,000 to 12,500 radiocarbon years BP and the
duration of calcite addition simulates the expected mean age of the added calcite is 18,750 radiocarbon years BP. Calcite
duration of sub-aerial exposure for samples of this age enrichment factor was assigned based on a worst case in our leaching
experiment. In each hypothetical sample, the effect of 60% weight loss
as determined from the Barbados sea level record by chemical leaching is assumed to enrich the initial fraction of calcite
(Fairbanks, 1989). We also assumed secondary calcite by a factor of three. Only samples with o0.2% calcite have apparent
14
has been added continuously into the coral during C ages that would be equal to the true ages (i.e. within 1s analytical
exposure and that the average 14C age of the calcite is error) in a 50–60% weight loss scenario. (b) Apparent 14C age as a
18,750 years BP. This assumption is based on the function of fractional weight loss and initial calcite percentage in a
hypothetical fossil coral. The ‘‘true’’ 14C age in each example is
progressive alteration of fossil corals sampled from assumed to be 45,000 years BP. In this scenario, we assume the reef
progressively older reef tracts (Mesolella et al., 1969; was uplifted at 45,000 years BP and has been sub-aerially exposed to
Bloom et al., 1974). For the simplicity of the model, we the present. Calcite has been added since the time of uplift and the
assume the less soluble calcite (Chave et al., 1962) is mean age of the added calcite is 22,500 radiocarbon years BP. Calcite
resistant to acid and only primary aragonite dissolves enrichment factor was assigned based on a worst case in our leaching
experiment. In each hypothetical sample, the effect of 60% weight loss
during a hypothetical pretreatment step prior to 14C by chemical leaching is assumed to enrich the initial fraction of calcite
dating. Finally, we assume the case that the calcite by a factor of three. Note only samples with o0.2% calcite have 14C
content is enriched by a factor of three (based on a worst apparent ages that would be equal to the true ages (i.e. within 1s
case in Table 2) after leaching to 60% weight loss, and analytical error) in a 50–60% weight loss scenario.
assign an enrichment factor to each fractional weight
loss to simulate the effect of leaching. in the coral and the weight fraction chemically leached.
The model output results in a series of calculated ages In the model case of a sample with a calcite concentra-
based on the initial amount of calcite (before leaching) tion of 1% that is dissolved to 50% of the original mass
 ARTICLE IN PRESS
1802 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

during the pretreatment, the measured age is 4200 48000

years younger than the true 14C age of the coral; for a

Apparent 230Th/234U/238U age (years BP)

sample having 3% initial calcite, the apparent 14C age 47900
can be
700 years younger for the same 50% weight loss
47800
(Fig. 3a).
In a second calcite addition model, we adopt the
47700
same assumptions and enrichment factors above but
consider a coral with a true 14C age of 45,000 years BP 47600
and which has had secondary calcite continuously
added to the present time. The average 14C age of 47500
the added calcite is 22,500 years BP in this scenario 0.2% calcite
0.5% calcite
(Fig. 3b). In the second model, a sample with a calcite 47400 1.0% calcite
2.0% calicte
concentration of 1% and dissolved to 50% of the 3.0% calicte
original mass during pretreatment acquires a measured 47300
age
2500 years younger than the true 14C age of the 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Assumed U concentration in seconday calcite (ppm)
coral. Greater than 2500-year offsets in 14C ages are
consistent with the outliers displayed in the range of Fig. 4. Apparent 230Th/234U/238U age as a function of the assumed U
40,000–50,000 years BP of the published coral data content of secondary calcite and the calcite percentage in a
hypothetical fossil coral. The true age of the primary aragonite is
(Fig. 1). A sample having 0.2% initial calcite acquires an
assumed to be
48,000 years BP (calendar age), and the average age of
apparent 14C age that is
600 years younger after 50% the calcite is assumed to be
24,000 years BP (calendar age). The
weight loss in this particular scenario. Though a 600- largest calculated age offset predicted by the 0.2% calcite scenario is
year offset can reasonably be considered to fall within about 30 years. Note we do not acid-leach fossil corals for
230
the analytical error of radiocarbon dating for an Th/234U/238U dated sub-samples and so there is no calcite
enrichment issue.

45,000 years BP sample, we conclude that only fossil
corals determined to have less than ‘‘0.2% calcite’’ can
be considered for accurate radiocarbon calibration Unlike 14C ages, the accuracies of corresponding
(Fairbanks et al., 2005). U-series dates are presumably not sensitive to low
We did not obtain radiocarbon ages for the samples levels of diagenetic calcite for several reasons. First, the
which were enriched in calcite during our etching tighter rhombohedral crystal structure of the calcite
experiments (Table 2), but we do have radiocarbon discriminates against the relatively large uranium atom
measurements on samples contaminated with varying (Milliman, 1974a) resulting in lower uranium concen-
amounts of secondary calcite. The results from pre- trations in calcite cements compared to coral. Second,
viously cited dissolution experiments (Burr et al., 1992; we do not acid-leach sub-samples for U-series dating
Yokoyama et al., 2000), in fact, appear to contradict our because the insolubility of thorium might potentially
model predictions. In both of those studies chemical lead to loss and re-absorption of 232Th and 230Th.
etching of up to 50% in calcite-bearing fossil corals Nevertheless, we can estimate age offsets for the
230
resulted in radiocarbon ages that were older than that of Th/234U/238U ages of coral samples by assuming a
the bulk sample. In Yokoyama (2000), radiocarbon uranium content of 3 ppm for coral aragonite (Cross
dating of progressively dissolved aliquots (following and Cross, 1983; Min et al., 1995) and 0.1–1.5 ppm for
50% etching) yielded 14C ages that agreed within the secondary calcite. In an extreme hypothetical case, a
analytical uncertainty. How and where secondary calcite 48,000 years BP (calendar age) coral sample, containing
precipitates within the coral aragonite matrix probably 3% calcite at 1.5 ppm U, acquires an apparent age
400
determines its susceptibility to selective dissolution. In years younger than the true age (Fig. 4). Therefore, the
the Yokoyama et al. (2000) study, secondary calcite was age offsets in 230Th/234U/238U ages of coral samples
found to fill coralline voids and attempts were then (screened with o0.2% calcite criteria) due to secondary
made to mechanically remove the phase. That the calcite are likely to be a small fraction of the reported
0–30% dissolved fraction yielded younger radiocarbon calendar age uncertainty.
ages suggests other sources of 14C contamination.
Younger or modern carbon may contaminate fossil 3.3. Organic carbon contamination
coral as either an organic or inorganic phase. An acid
leach (as phosphoric or HCl) would likely be very Corals from locations such as Papua New Guinea
effective in removing any surface bound contaminants. (Edwards et al., 1993; Yokoyama et al., 2000; Cutler
However, depending on the specifics of the chemical et al., 2004), Barbados, Santo and Araki Islands
etch, progressive leaching beyond 30% dissolution may (Fairbanks et al., 2005) may have been exposed to a
or may not produce uniform dissolution of calcite and freshwater environment during portions of their history
aragonite, due to uplift and/or lowering of sea level allowing soil,
 ARTICLE IN PRESS
T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808 1803

plant roots or microbes the opportunity to penetrate similar effects of different pretreatments on the 14C
and spread in fossil coral outcrops. Corals may be concentration detected in 4100,000 years old foramini-
contaminated with mold during collection, storage and fera tests. In that study, ultrasonic and acid leaching
shipment under humid condition. Modern carbon may (either using HCl or 30% H2O2) was applied. The lowest
14
also be introduced during sample handling. Some C blank values were achieved by the combination of a
fraction of the organic matter is fairly resistant to H2O2 leach and elimination of any dry-down step. Since
oxidation as evident by progressive whitening of fossil we know of no a priori screening technique that is
coral samples over several days of continuous soaking in sensitive enough to quickly identify samples contami-
30% H2O2. Based on this qualitative visual assessment, nated with trace amounts of resistant organic carbon,
we tested the efficacy of a 7-day H2O2 treatment. Seven we adopted the conservative and somewhat time-
days were an arbitrary doubling of the 3 days we consuming step of treating all samples older than
observed progressive whitening of corals soaked in 30% 30,000 years BP with 7-day H2O2 ultrasonic oxidation
H2O2 and is reaching practical time limitations. treatment. However, in an attempt to further eliminate
We tested the extended (7-day) H2O2 cleaning techni- the possibility of contamination from resistant organic
que for corals older than 30,000 years BP for two reasons. carbon, we subsequently modified the H2O2 treatment
First, these samples were exposed to sub-aerial conditions from that used in Table 3 so that the ultrasonic bath
including exposure to soils and other organic contami- procedure consists of a continuous 1-h on/off cycle for 7
nants. Second, radiocarbon ages of older samples are days. Based on Schleicher et al.’s (1998) results, we have
much more sensitive to modern 14C contamination. avoided any dry-down step in the sample preparation of
Coral samples were handled with gloves and were coral sub-samples for 14C dating. Using these cleaning
ultrasonically cleaned in 30% H2O2 using a high power techniques, the background 14C levels measured in Kiel
ultrasonic probe for 2 h. Samples were then transferred on a 98,000-year-old Araki coral (AK-H-2) ranged from
into 5 ml glass vials, filled with 30% H2O2 and placed in 0.12% to 0.15% modern carbon, equivalent to an
a sonication bath (BRANSON 5210) for 1 h. The average radiocarbon age of 53,000 years BP.
samples remained in the 30% H2O2 for 24 h after which
the H2O2 was replaced and the samples were sonicated 3.4. 26,000– 50,000 years BP calibration
for 1 h. The process was repeated for a total of 7 days.
Finally, the samples were placed in fresh 30% H2O2 and In an effort to extend the radiocarbon calibration
sonicated with a high power probe for 2 h. Samples were beyond 26,000 years BP, we applied our XRD screening
immediately placed and stored (damp) in glass vials to more than 100 fossil coral samples in this study.
without drying in order to avoid any mineral precipita- Based on terrace location, we selected only Araki
tion. Sub-samples used for uranium-series dating were samples that were expected to fall in the range of
not subject to the H2O2 treatment. radiocarbon dating for screening. The majority of the
Paired radiocarbon dates were obtained from three samples were either collected from outcrops at Araki
coral samples (with and without the 7-day H2O2 Island during a field expedition in 2004 or were sampled
oxidation treatment) described above. Two samples from the existing collection of A. Bloom, at Cornell
showed no effect of the extended H2O2 treatment based University. The Barbados samples (RGF) were collected
on 14C ages that overlap at 1s analytical uncertainty during offshore drilling of Barbados submerged reef
(Table 3). For the remaining sample, AK-BD-3, the terraces (Fairbanks, 1989). One-third of the samples
H2O2-treated sample is several hundred years older than passed the ‘‘o0.2% calcite’’ criteria by our XRD
the untreated companion sample. The data suggest that methods and were therefore subjected to the extended
some samples are free of oxidation-resistant organic hydrogen peroxide cleaning prior to processing for 14C
matter while an occasional sample appears to be dating. U-series ages were determined without any
contaminated. Schleicher et al. (1998) demonstrated leaching pretreatment. The resulting coral radiocarbon

Table 3
14
Comparison between C ages of hydrogen peroxide treated versus untreated fossil corals
14 14 14
Sample Non-treated C age (uncorrected) H2O2-treated C age (uncorrected) C age offset (years)
(years BP) (years BP)

AK-BD-3 29,0907190 29,7007220 610

AK-BD-2 30,6907210 30,3607240 330
AK-BD-11 32,9007300 33,3907350 490

Note: All samples are from Araki Island and contain o0.2% calcite. Errors represent 1s analytical uncertainties. All 14C dates in this table were
provided by the Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratory. All 14C dates in this table were not
corrected for reservoir age, which is different from Table 4.
 ARTICLE IN PRESS
1804 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

14
calibration data (U series and AMS C) are listed in (Guyodo and Valet, 1996; Guyodo and Valet, 1999; Laj
Table 4 and displayed in Fig. 5. et al., 2000, 2004), and recent published K–Ar and
40
Ar/39Ar dates (40,40072,000 years BP) for the type
locality (Guillou et al., 2004) refines the timing of
4. Discussion Laschamp event. The most distinct feature of our coral
data is the general trend of increasing offset between 14C
The significant scatter displayed by previously pub- ages and calendar ages; however, there is no significant
lished coral radiocarbon calibration data older than anomaly between 38,000 and 42,000 years BP (Fig. 5).
30,000 years BP (Yokoyama et al., 2000) suggests that Since our coral data in this study (26,000–50,000 years
the accuracy of the 14C ages of these samples may not be BP) were obtained mainly from outcrop samples, which
well constrained (Fig. 1) and that stepwise dissolution are not stratigraphically continuous, it is possible that
may not have been sufficient for identifying potential we simply did not sample corals that date to the
contamination from all forms of younger carbon. It is Laschamp excursion. Alternatively, the ‘‘duration’’ of
inevitable that the scarcity of unaltered samples in this Laschamp excursion and 14C production spike may not
age range contribute to selection of less than ideal be long enough to be recorded in surface ocean and
quality samples and the identification of measurable fossil corals (Beck et al., 2001; Hughen et al., 2004).
amounts of calcite in many of these samples (Yokoyama Similarly, although the ‘‘magnitude’’ of Laschamp
et al., 2000) indicates the potential for diagenetic related excursion seems significant on the scale of paleointensity
artifacts. As we progressed in our first attempts to in most stacks, the resulting carbon cycle models of
radiocarbon date fossil coral older than 30,000 years BP global 14C indicate that the Laschamp might not be
we clearly identified unacceptable scatter (Fig. 5) in the large enough to be recorded in the fossil corals. The lack
results, even though most of these samples passed the of a prominent Laschamp anomaly in our coral data
established criteria (Reimer et al., 2002) for selection might also be due to the non-linearity relationship
(d234U, negligible 232Th and o1% calcite). We have between geomagnetic intensity and global 14C produc-
argued in the previous sections that more stringent tion if the intensity dropped more than 20% of its
sample selection criteria and pretreatment steps are now present value when decreasing intensity has little effect
required. We now present data to strongly suggest that on global 14C production (Elsasser et al., 1956; Lal,
our sample screening and pretreatment steps have 1988).
eliminated the most common sources of radiocarbon
contamination and provide more accurate data than
previously published (Fig. 5). We base this conclusion
on several factors. First, our data yield a relatively 5. Conclusions
smooth, continuous curve with little ‘‘noise’’ from point
to point, in contrasts sharply with the data of The accuracy and precision of the radiocarbon
Yokoyama et al. (2000) that yield seemingly impossible calibration by fossil corals are ultimately limited by
rapid and large 14C production or carbon cycle mixing the availability and quality of samples. Published coral
14
scenarios (Oeschger et al., 1975; Lal, 1988; Mikolaje- C dates for samples older than 30,000 calendar years
wicz, 1996). Second, our data are consistent with the BP (Bard et al., 1998; Yokoyama et al., 2000; Cutler et
general trend of reconstructed geomagnetic paleointen- al., 2004) are, however, too disparate for practical
sity stacks. These stacks demonstrate that the geomag- calibration. A large portion of the published coral data
netic field intensity fell from the highest values to the in the age range of 30,000–50,000 years BP (Yokoyama
lowest between 50,000 and 40,000 years BP (Guyodo et al., 2000) were generated from samples containing
and Valet, 1996; Guyodo and Valet, 1999; Laj et al., diagenetic calcite (42%). We present a simple model
2000, 2004). The decrease would have reduced the which demonstrates that contamination of coral with
geomagnetic shielding effect and resulted in increased the amounts of secondary calcite commonly reported in
global 14C production (Elsasser et al., 1956; Lal, 1988). the literature (1–3%) may explain at least part of the
Excess 14C in the atmosphere contributes to larger observed offsets and scatter in published coral data.
offsets between 14C ages and calendar ages if recorded, Model calculations demonstrate that significant dating
and we observe such a trend during the time period artifacts can be generated in samples containing greater
50,000–40,000 years BP in our coral data (Fig. 5). Our than 0.5% calcite. Contamination by calcite may be
data also displays a trend more consistent with the increased during 14C graphite target preparation when
modeled 14C concentrations based on geomagnetic field samples are subjected to an acid leach and subsequent
intensity (Laj et al., 2002; Beck et al., 2001; Hughen weight loss (50–60%) prior to dating, during which the
et al., 2004). fractional mass of calcite may increase by a factor of two
All of the paleointensity stacks show a distinctive or three. We demonstrate our ability to detect less than
excursion, Laschamp, around 38,000–42,000 years BP 0.2% calcite in aragonite by XRD and we apply
Table 4
230
Th/234U/238U ages and 14
C ages of Barbados and Araki fossil corals in this study
232
Sample sp. U-series [U] 1s Th [234U/238U] 1s d234Uinitial 1s [230Th/234U] 1s Th/U age 1s 14
C lab ID 14
C ageRC 1s
ID (ppm) (pg/g) (years BP) (years BP)

RGF 12-28-6 Ap 121801TC1 3.1239 0.0008 365 1.1298 0.0005 141.1 0.7 0.2388 0.0003 29,590 44 CAMS 73919 24,925 60
RGF 12-28-6 CAMS 73953 24,975 70
RGF 12-28-7 Ap 021301RM2 2.9820 0.0005 482 1.1274 0.0004 138.7 0.5 0.2432 0.0004 30,214 56 CAMS 73920 24,955 60
RGF 12-28-7 073101RM2 3.2159 0.0008 863 1.1284 0.0005 139.9 0.5 0.2423 0.0004 30,080 58 CAMS 73949 24,815 60
RGF 12-28-7 CAMS 75385 24,845 90
RGF 12-28-7 CAMS 77432 24,915 90
RGF 12-29-2 Ap CAMS 73921 24,955 70
RGF 12-29-2 080201RM2 3.2406 0.0007 68 1.1269 0.0004 138.3 0.5 0.2433 0.0003 30,225 40 CAMS 73946 24,915 70
AK-BD-2 – 102803TC5 1.9947 0.0007 10 1.1263 0.0006 139.9 0.9 0.2829 0.0003 36,014 43 CAMS 99632 30,325 210
AK-BD-2 CAMS 104452 29,995 240
AK-BD-3 P 102903TC1 3.9029 0.0013 8 1.1268 0.0005 139.8 0.8 0.2725 0.0003 34,467 39 CAMS 99022 28,725 190

T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

AK-BD-3 CAMS 104453 29,335 220
AK-BD-4 P 102903TC2 2.7232 0.0007 29 1.1274 0.0005 140.9 0.8 0.2796 0.0003 35,518 43 CAMS 99633 30,745 220
AK-BD-5 P 062603TC2 2.6362 0.0009 13 1.1264 0.0005 140.1 0.8 0.2844 0.0005 36,247 74 CAMS 99015 31,305 260
AK-BD-9 Pl 102903TC6 2.8262 0.0008 734 1.1249 0.0005 139.8 0.8 0.3074 0.0003 39,759 42 CAMS 99024 34,725 400
AK-BD-10 Fs 010803RM1 2.5785 0.0005 12 1.1263 0.0004 141.3 0.6 0.3069 0.0003 39,684 50 CAMS 93343 34,725 430
AK-BD-10 CAMS 99637 35,045 370
AK-BD-11 Pl 063003TC3 2.6913 0.0006 299 1.1256 0.0004 140.0 0.8 0.2986 0.0004 38,401 58 CAMS 99016 32,535 300
AK-BD-11 CAMS 104454 33,025 350
AK-L-1 Fs 010803RM2 2.1286 0.0004 16 1.1310 0.0004 144.1 1.1 0.2672 0.0003 33,677 39 CAMS 93342 29,365 230

ARTICLE IN PRESS
AK-L-1 CAMS 99636 29,095 180
AK-K-1 P 112403TC1 3.8042 0.0011 27 1.1273 0.0007 141.0 1.1 0.2844 0.0003 36,240 46 CAMS 104459 31,775 300
AK-K-1 CAMS 106898 31,765 490
AK-K-2 Plla 042203TC2 2.4537 0.0005 24 1.1279 0.0005 141.7 1.5 0.2842 0.0003 36,216 49 KIA 24977 30,615 150
AK-F-1 F 082504TC1 2.8052 0.0008 13 1.1276 0.0007 141.3 1.0 0.2834 0.0004 36,091 67 CAMS 106896 31,885 500
AK-F-1 CAMS 106893 32,255 530
AK-F-2 G 082504TC3 2.5551 0.0006 9 1.1287 0.0006 142.5 0.9 0.2831 0.0006 36,047 96 CAMS 106897 31,245 460
AK-AA-1 Fs 082504TC4 2.2705 0.0005 15 1.1280 0.0004 142.5 0.7 0.2942 0.0004 37,724 68 CAMS 106890 33,725 630
AK-BE-1b Pl 063003TC4 2.7054 0.0006 337 1.1228 0.0005 139.7 0.7 0.3446 0.0006 45,689 93 CAMS 99018 41,945 960
AK-BE-1b 082604TC3 2.6771 0.0007 300 1.1252 0.0006 142.5 0.9 0.3446 0.0006 45,681 98
AK-AB-1 Pl 042303TC1 2.5163 0.0005 57 1.1231 0.0005 141.1 0.7 0.3602 0.0003 48,273 49 KIA 24976 44,045 530
AK-AB-1 111803TC4 2.7484 0.0007 70 1.1226 0.0005 140.5 0.8 0.3595 0.0003 48,164 45
ARA04-2D G 112304RM1 2.3337 0.0004 29 1.1299 0.0005 145.3 0.8 0.3077 0.0009 39,799 140 KIA 24978 34,685 230
ARA04-10D F 112304TC2 2.4515 0.0004 9 1.1299 0.0005 145.4 0.7 0.3085 0.0004 39,930 58 KIA 24979 34,205 210
ARA04-11D G 102604LC5 2.4407 0.0011 36 1.1280 0.0005 143.7 0.8 0.3149 0.0003 40,923 54 KIA 24980 35,275 200
ARA04-17D — 011805TC1 3.1012 0.0007 6 1.1289 0.0005 142.9 0.8 0.2863 0.0003 36,521 44 KIA 24981 31,345 150
ARA04-19D P 011805TC2 3.3744 0.0009 94 1.1307 0.0005 144.7 0.8 0.2830 0.0003 36,031 44 KIA 24982 31,705 150
ARA04-22D P 112304RM4 2.9510 0.0005 44 1.1323 0.0005 146.5 0.7 0.2829 0.0004 36,008 64 KIA 24984 30,855 140
ARA04-40E Pl 102604RM4 2.4933 0.0005 10 1.1236 0.0005 142.1 0.7 0.3658 0.0003 49,203 45 KIA 24987 44,795 570
ARA04-43E P 092204LC2 2.8368 0.0007 17 1.1252 0.0005 143.5 0.7 0.3597 0.0006 48,183 102 KIA 24988 45,695 630
ARA04-44E P 092204LC3 3.1435 0.0011 69 1.1262 0.0006 144.7 0.9 0.3609 0.0007 48,378 112 KIA 24989 44,385 550
ARA04-71E F 092304LC1 2.2226 0.0008 27 1.1258 0.0005 142.2 0.8 0.3312 0.0005 43,519 82 KIA 24991 38,495 380
ARA04-73E – 092304LC2 2.1453 0.0005 14 1.1270 0.0005 143.9 0.8 0.3348 0.0005 44,098 83 KIA 24992 39,345 310
ARA04-75E P 092304LC3 2.9473 0.0005 27 1.1254 0.0005 142.3 0.7 0.3395 0.0006 44,850 90 KIA 24993 41,345 390
ARA04-77E – 092304LC4 2.6076 0.0005 8 1.1280 0.0005 144.9 0.8 0.3333 0.0005 43,840 86 KIA 24994 39,335 320
ARA04-85E G 092304LC5 2.5620 0.0005 4 1.1262 0.0004 142.8 0.7 0.3310 0.0004 43,474 72 KIA 24995 40,095 330

Note: Coral species: ‘‘Ap’’ is Acropora palmata and ‘‘A’’ is Acropora species. ‘‘Pl’’ is Porites lutea and ‘‘P’’ is Porites species. ‘‘Fs’’ is Favia stelligera and ‘‘F’’ is Favia species. ‘‘G’’ is Goniastrea. ‘‘Plla’’
is Platygyra lamellina and ‘‘Pl’’ is Platygyra. ‘‘—’’ Represents unidentified species. ‘‘[ ]’’ Represents activity ratio. All U-series analyses were made by an MC-MS-ICPMS (Plasma 54) at LDEO
(Halliday et al., 1995; Mortlock et al., 2005). 230Th/234U/238U ages are represented by ‘‘Th/U ages’’ in this table and were calculated using the following equation (Ivanovich et al., 1992):
 
1  el230 t1 1 l230
½230 Th=234 U ¼ þ 1  ð1  eðl230 l234 Þt1 Þ.
½234 U=238 U ½234 U=238 U l230  l234

In all calculations, we use decay constants of l230 ¼ 9:1577  106 year1 and l234 ¼ 2:8263  106 year1 (Cheng et al., 2000). 14C ageRC reported in this table are based on the definition in Stuiver
and Polach (1977) and have been corrected for reservoir age of 365 years, which was determined by differencing Holocene coral data from Santo, Araki and Barbados (Fairbanks et al., 2005), and

1805
other data from Vanuatu (Paterne et al, 2004) from the tree-ring calibration data set (Reimer et al., 2004). ‘‘CAMS’’ represents 14C dates performed at Lawrence Livermore National Laboratory,
University of California. ‘‘KIA’’ represents 14C dates performed in the Leibniz Laboratory for Dating and Isotope Analyses, Christian-Albrechts-Universität, Kiel, Germany.
 ARTICLE IN PRESS
1806 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

Papua New Guinea (Yokoyama et al. 2000)

New Guinea (Bard et al., 1998)
Dating and Isotope Analyses, Christian-Albrechts-Uni-
50000
Santo and Papua New Guinea (Cutler et al., 2004) versität, Kiel, Germany under the supervision of P.M.
Araki and Barbados corals with >0.2% calcite (this study)
Araki and Barbados corals with <0.2% clacite (this study) Grootes. P. Biscaye provided technical support and
Equiline
advice on lowering the calcite detection limit at the
45000
XRD facility at LDEO. L. Britt assisted with developing
Conventional 14C age (years BP)

the XRD methodology. L. Cao assisted with the 14C

40000
sample preparation and U-series and XRD measure-
ments. T. Fairbanks, M. Tamata, J. Tamata, J. Rongo
35000 and A. Somoli collected additional samples during a
2004 expedition to Araki Island. We thank two
30000 anonymous reviewers for their helpful comments. This
work was supported by grants from the National
25000 Science Foundation (OCE99-11637, ATM03-27722)
and the Climate Center at LDEO (6-80303). This is
20000
Lamont Contribution 6765.
30000 35000 40000 45000 50000
230
Th/234U/238U age (years BP)
References
Fig. 5. Published radiocarbon and U-series ages for fossil corals
between 26,000 and 50,000 years BP (Bard et al., 1998; Yokoyama et Bard, E., Hamelin, B., Fairbanks, R.G., Zindler, A., 1990a. Calibra-
al., 2000; Cutler et al., 2004) compared to our data from Barbados and tion of the 14C timescale over the past 30,000 years using mass
Araki Islands (Table 4). The equiline is presented only for reference. spectrometric U-Th ages from Barbados corals. Nature 345,
Yokoyama et al. (2000) reported calcite concentrations in the range of 405–410.
0–3%, Bard et al. (1998) reported o1% and Cutler et al. (2004) Bard, E., Hamelin, B., Fairbanks, R.G., Zindler, A., Mathieu, G.,
reported o1%. Barbados and Araki coral samples were a priori Arnold, M., 1990b. U/Th and 14C ages of corals from Barbados
screened by XRD method described in this study and were determined and their use for calibrating the 14C time scale beyond 9000 years
to have o0.2% calcite. B.P. Nuclear Instruments and Methods in Physics Research B 52,
461–468.
Bard, E., Arnold, M., Fairbanks, R.G., Hamelin, B., 1993. 230Th/234U
‘‘o0.2% calcite’’ as our a priori criterion to select coral and 14C ages obtained by mass spectrometry on corals. Radio-
carbon 35 (1), 191–199.
samples. Bard, E., Arnold, M., Hamelin, B., Tisnerat-Laborde, N., Cabioch,
Organic matter is another source of radiocarbon G., 1998. Radiocarbon calibration by means of mass spectrometric
230
contamination and is particularly common in outcrop Th/234U and 14C ages of corals: an updated database including
coral samples exposed to soils or in corals having mold samples from Barbados, Mururoa and Tahiti. Radiocarbon 40 (3),
colonization in transit or during storage after sampling 1085–1092.
Beck, J.W., Richards, D.A., Edwards, R.L., Silverman, B.W., Smart,
in humid environments. We observe visible ‘‘whitening’’ P.L., Donahue, D.J., Hererra-Osterheld, S., Burr, G.S., Calsoyas,
of typical outcrop samples following ultrasonic cleaning L., Jull, A.J.T., Biddulph, D., 2001. Extremely large variations of
in 30% H2O2 over a several day period that we attribute atmospheric 14C concentration during the last glacial period.
to continued oxidation of organic matter. An extended Science 292, 2453–2458.
7-day hydrogen peroxide pretreatment is used to remove Bloom, A.L., Broecker, W.S., Chappell, J.M.A., Matthews, R.K.,
Mesolella, K.J., 1974. Quaternary sea level fluctuations on a
any contaminating organic matter. We demonstrate that tectonic coast: new 230Th/234U dates from the Huon Peninsula,
cleaning in hydrogen peroxide yields older or reprodu- New Guinea. Quaternary Research 4 (2), 185–205.
cible 14C dates compared to non-treated samples. These Burr, G.S., Edwards, R.L., Donahue, D.J., Druffel, E.R.M., Taylor,
are more rigorous screening and cleaning techniques F.W., 1992. Mass spectrometric 14C and U-Th measurements in
coral. Radiocarbon 34 (3), 611–618.
than those adopted by IntCal04 (Reimer et al., 2002) for
Burr, G.S., Beck, J.W., Taylor, F.W., Recy, J., Edwards, R.L.,
samples younger than 26,000 years BP. However, for Cabioch, G., Correge, T., Donahue, D.J., O’Malley, J.M., 1998. A
samples older than 26,000 years BP these screening and high-resolution radiocarbon calibration between 11,700 and 12,400
cleaning techniques become essential and provide calendar years BP derived from 230Th ages of corals from Espiritu
precise, reproducible and accurate radiocarbon calibra- Santo Island, Vanuatu. Radiocarbon 40 (3), 1093–1105.
tion data. Chave, K.E., Deffeyes, K.S., Weyl, P.K., Garrels, R.M., Thompson,
M.E., 1962. Observations on the solubility of skeletal carbonates in
aqueous solutions. Science 137, 33–34.
Cheng, H., Edwards, R.L., Hoff, J., Gallup, C.D., Richards, D.A.,
Acknowledgements Asmerom, Y., 2000. The half-lives of uranium-234 and thorium-
230. Chemical Geology 169, 17–33.
Radiocarbon measurements were performed at the Cross, T.S., Cross, B.W., 1983. U, Sr, and Mg in Holocene and
Pleistocene corals A. palmata and M. annularis. Journal of
Center of Accelerator Mass Spectrometry at Lawrence Sedimentary Petrology 53, 587–594.
Livermore National Laboratory, University of Califor- Cutler, K.B., Gray, S.C., Burr, G.S., Edwards, R.L., Taylor, F.W.,
nia by T. Guilderson and at the Leibniz Laboratory for Cabioch, G., Beck, J.W., Cheng, H., Moore, J., 2004. Radiocarbon
 ARTICLE IN PRESS
T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808 1807

calibration and comparison to 50 kyr BP with paired 14C and 230Th Kitagawa, H., van der Plicht, J., 2000. Atmospheric radiocarbon
dating of corals from Vanuatu and Papua New Guinea. Radio- calibration beyond 11,900 cal B.P. from Lake Suigetsu laminated
carbon 46 (3), 1127–1160. sediments. Radiocarbon 42, 369–380.
Damon, P.E., Lerman, J.C., Long, A., 1978. Temporal fluctuations of Lal, D., 1988. Theoretically expected variations in the terrestrial cosmic-
atmospheric 14C: causal factors and implications. Annual Review ray production rates of isotope. In: Castagnoli, G.C. (Ed.),
of Earth and Planetary Sciences 6, 457–494. Solar–Terrestrial Relationships and the Earth Environment in the
de Vries, H.L., 1958. Variation in concentration of radiocarbon Last Millennia. North-Holland, Amsterdam, New York, pp. 216–233.
with time and location on Earth. Proceedings Koninklijke Laj, C., Kissel, C., Mazaud, A., Channell, J.E.T., Beer, J., 2000. North
Nederlandse. Akademie van Wetenschappen, Series B 61 (2), Atlantic palaeointensity stack since 75 ka (NAPIS-75) and the
94–102. duration of the Laschamp event. Philosophical Transactions of the
Edwards, R.L., Beck, J.W., Burr, G.S., Donahue, D.J., Chappell, Royal Society of London, Series A, Mathematical Physical and
J.M.A., Bloom, A.L., Druffel, E.R.M., Taylor, F.W., 1993. A large Engineering Sciences 358, 1009–1025.
drop in atmospheric 14C/12C and reduced melting in the Younger Laj, C., Kissel, C., Mazaud, A., Michel, E., Muscheler, R., Beer, J.,
Dryas, documented with 230Th ages of corals. Science 260, 2002. Geomagnetic field intensity, North Atlantic deep water
962–968. circulation and atmospheric D14C during the last 50 kyr. Earth and
Edwards, R.L., Cheng, H., Murrell, M.T., Goldstein, S.J., 1997. Planetary Science Reviews 200, 177–190.
Protactinium-231 dating of carbonates by thermal ionization mass Laj, C., Kissel, C., Beer, J., 2004. High resolution global paleointensity
spectrometry: implications for Quaternary climate change. Science stack since 75 kyr (GLOPIS-75) calibrated to absolute values. In:
276, 782–786. Channell, J.E.T., Kent, D.V., Lowrie, W., Meert, J.G. (Eds.),
Elsasser, W., Ney, E.P., Winckler, J.R., 1956. Cosmic-ray intensity and Timescales of the Palemagnetic Field, AGU Geophysical Mono-
geomagnetism. Nature 178, 1226–1227. graph 145, pp. 255–265.
Fairbanks, R.G., 1989. A 17,000-year glacio-eustatic sea level record: Mesolella, K.J., Matthews, R.K., Broecker, W.S., Thurber, D.L.,
influence of glacial melting rates on the Younger Dryas event and 1969. The astronomical theory of climatic change: Barbados data.
deep-ocean circulation. Nature 342, 637–642. Journal of Geology 77 (3), 250–274.
Fairbanks, R.G., Mortlock, R.A., Chiu, T.-C., Cao, L., Kaplan, A., Mikolajewicz, U., 1996. A meltwater induced collapse of the ‘conveyor
Guilderson, T.P., Fairbanks, T.W., Bloom, A.L., Grootes, P.M., belt’ thermohaline circulation and its influence on the distribution
Nadeau, M.-J., 2005. Radiocarbon calibration curve spanning 0 to of D14C and d18O in the oceans. Max-Planck-Institut fur
50,000 years BP based on paired 230Th/234U/238U and 14C dates on Meteortologie Report 189.
pristine corals. Quaternary Science Reviews, this issue, doi:10.1016/ Milliman, J.D., 1974a. Carbonates and the ocean. In: Milliman, J.D.
j.quascirev.2005.04.007. (Ed.), Recent Sedimentary Carbonates 1, Marine Carbonates.
Gill, I., Olson, J.J., Hubbard, D.K., 1995. Corals, paleotemperature Springer, Berlin, pp. 3–15.
records, and the aragonite–calcite transformation. Geology 23, Milliman, J.D., 1974b. Methods. In: Milliman, J.D. (Ed.), Recent
333–336. Sedimentary Carbonates 1, Marine Carbonates. Springer, Berlin,
Goslar, T., Arnold, M., Tisnerat-Laborde, N., Czernik, J., Wieck- pp. 16–33.
owski, K., 2000. Variations of Younger Dryas atmospheric Min, G.R., Edwards, R.L., Taylor, F.W., Recy, J., Gallup, C.D., Beck,
radiocarbon explicable without ocean circulation changes. Nature J.W., 1995. Annual cycles of U/Ca in coral skeletons and U/Ca
403, 877–880. thermometry. Geochimica et Cosmochimica Acta 59, 2025–2042.
Guillou, H., Singer, B.S., Laj, C., Kissel, C., Scaillet, S., Jicha, B.R., Mortlock, R.A., Fairbanks, R.G., Chiu, T.-C., Rubenstone, J., 2005.
230
2004. On the age of the Laschamp geomagnetic excursion. Earth Th/234U/238U and 231Pa/235U ages from a single fossil coral
and Planetary Science Letters 227, 331–343. fragment by multi-collector magnetic-sector inductively coupled
Guyodo, Y., Valet, J.-P., 1996. Relative variations in geomagnetic plasma mass spectrometry. Geochimica et Cosmochimica Acta 69,
intensity from sedimentary records: the past 200,000 years. Earth 649–657.
and Planetary Science Letters 143, 23–36. Nadeau, M.-J., Grootes, P.M., Voelker, A., Bruhn, F., Duhr, A.,
Guyodo, Y., Valet, J.-P., 1999. Global changes in intensity of the Oriwall, A., 2001. Carbonate 14C background: does it have
Earth’s magnetic field during the past 800 kyr. Nature 399, multiple personalities? Radiocarbon 43, 169–176.
249–252. Oeschger, H., Siegenthaler, U., Schotterer, U., Gugelmann, A., 1975.
Halliday, A.N., Lee, D.-C., Christensen, J.N., Walder, A.J., Freeman, A box diffusion model to study carbon dioxide exchange in nature.
P.A., Jones, C.E., Hall, C.M., Yi, W., Teagle, D., 1995. Recent Tellus 27 (2), 168–192.
developments in inductively coupled plasma magnetic sector Paterne, M., Ayliffe, L.K., Arnold, M., Cabioch, G., Tisnerat-
multiple collect mass spectrometry. International Journal of Mass Laborde, N., Hatte, C., Douville, E., Bard, E., 2004. Paired 14C
Spectrometry and Ion Processes 146/147, 21–33. and 230Th/U dating of surface corals from the Marquesas and
Hughen, K.A., Overpeck, J.T., Lehman, S.J., Kashgarian, M., Vanuatu (sub-equatorial Pacific) in the 3000 to 15,000 Cal yr
Southon, J., Peterson, L.C., Alley, R., Sigman, D.M., 1998. interval. Radiocarbon 46 (2), 551–566.
Delacial changes in ocean circulation from an extended radio- Reimer, P.J., Hughen, K.A., Guilderson, T.P., McCormac, G., Baillie,
carbon calibration. Nature 391, 65–68. M.G.L., Bard, E., Barratt, P., Beck, J.W., Buck, C.E., Damon, P.E.,
Hughen, K.A., Southon, J.R., Lehman, S.J., Overpeck, J.T., 2000. Friedrich, M., Kromer, B., Ramsey, C.B., Reimer, R.W., Remmele,
Synchronous radiocarbon and climate shifts during the last S., Southon, J.R., Stuiver, M., van der Plicht, J., 2002. Preliminary
deglaciation. Science 290, 1951–1954. report of the first workshop of the IntCal04 radiocarbon calibration/
Hughen, K., Lehman, S., Southon, J., Overpeck, J., Marchal, O., comparison working group. Radiocarbon 44 (3), 653–661.
Herring, C., Turnbull, J., 2004. 14C activity and global carbon Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W.,
cycle changes over the past 50,000 years. Science 303 (5655), Bertrand, C.J.H., Blackwell, P.G., Buck, C.E., Burr, G.S., Cutler,
202–207. K.B., Damon, P.E., Edwards, R.L., Fairbanks, R.G., Friedrich,
Ivanovich, M., Latham, A.G., Ku, T.-L., 1992. Uranium-series M., Guilderson, T.P., Hogg, A.G., Hughen, K.A., Kromer, B.,
disequilibrium applications in geochronology. In: Uranium-Series McCormac, G., Manning, S., Ramsey, C.B., Reimer, R.W.,
Disequilibrium: Applications to Earth, Marine, and Environmental Remmele, S., Southon, J.R., Stuiver, M., Talamo, S., Taylor,
Sciences, second ed., pp. 62–94. F.W., van der Plicht, J., Weyhenmeyer, C.E., 2004. Intcal04
 ARTICLE IN PRESS
1808 T.-C. Chiu et al. / Quaternary Science Reviews 24 (2005) 1797–1808

terrestrial radiocarbon age calibration, 0–26 cal kyr BP. Radio- Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen,
carbon 46, 1029–1058. K.A., Kromer, B., McCormac, G., van der Plicht, J., Spurk, M.,
Schleicher, M., Grootes, P.M., Nadeau, M.-J., Schoon, A., 1998. The 1998. INTCAL98 radiocarbon age calibration, 24,000–0 cal BP.
carbonate 14C background and its components at the Leibniz AMS Radiocarbon 40, 1041–1083.
facility. Radiocarbon 40 (1), 85–93. Suess, H.E., 1970. The three causes of the secular 14C fluctuations,
Stuiver, M., 1961. Variations in radiocarbon concentration and their amplitudes and time constants. In: Olsson, I.U. (Ed.),
sunspot activity. Journal of Geophysical Research 66, 273–276. Radiocarbon Variations and Absolute Chronology. Wiley Inter-
Stuiver, M., Polach, H., 1977. Discussion—Reporting of 14C data. science Division, New York, pp. 595–605.
Radiocarbon 19, 355–363. Voelker, A.H.L., Grootes, P.M., Nadeau, M.J., Sarnthein, M., 2000.
Stuiver, M., Quay, P.D., 1980. Changes in atmospheric carbon-14 Radiocarbon levels in the Iceland Sea from 25–53 kyr and their
attributed to a variable Sun. Science 207, 11–19. link to the Earth’s magnetic field intensity. Radiocarbon 42,
Stuiver, M., Kromer, B., Becker, B., Ferguson, C.W., 1986. Radio- 437–452.
carbon age calibration back to 13,300 years BP and the 14C age Yokoyama, Y., Esat, T.M., Lambeck, K., Fifield, L.K., 2000. Last ice
matching of the German Oak and US bristlecone pine chronolo- age millennial scale climate changes recorded in Huon Peninsula
gies. Radiocarbon 28 (2B), 969–979. corals. Radiocarbon 42 (3), 383–401.