RADIOCARBON, Vol 52, Nr 4, 2010, p 1639–1644 © 2010 by the Arizona Board of Regents on behalf of the University of Arizona
1639
CALIBRATION OF MANGERUD’S BOUNDARIES
Adam Walanus
Department of Geoinformatics, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Kraków, Poland.
Corresponding author. Email: walanus@geol.agh.edu.pl.
Dorota Nalepka
W. Szafer Institute of Botany, Polish Academy of Sciences, Lubicz 46, 31-512 Kraków, Poland.
ABSTRACT. The “calibration” of arbitrarily defined (in some sense, “conventional”) ages, given in conventional radiocarbon
years BP, is now becoming necessary because the term “radiocarbon age” is used less often in archaeological and Quaternary
practice. The standard calibration procedure is inappropriate here because Mangerud’s boundaries are not
measurement results. Thus, another approach to the problem is proposed in order to model the natural situation of many, uniformly
distributed, dated samples, which should be similarly divided by the original and “calibrated” boundary. However, the
result depends on the value of the typical measurement error and is not unequivocal.
THE PROBLEM
From the early (precalibration) development era of the radiocarbon method, some well-established
“dates” of global or regional events have existed in the literature. These “time horizons” summarize
a considerable number of 14C dates as well as lithological considerations. Mangerud’s chronozone
boundaries are probably the most important (Mangerud et al. 1974). Dates proposed as a summary
of the cumulated knowledge on climate change in the Holocene and Late Glacial epochs have the
sense of convention. But in fact, these are arbitrarily established dates, or “boundaries,” to which
individually recorded events should be referred. The notion of a “chronozone” is correct for the climate
stage while its boundaries are set as the absolute age. However, when one considers the date of
publication (1974) it is clear that the boundaries are given in 14C years (Conv BP). A problem arises
because Mangerud’s chronozones have been used extensively up to the present day (Webb 2006;
Anderson et al. 2007; Berglund et al. 2008; Haynes 2008; RybníËek and RybníËková 2008; Sørensen
2008; Wanner et al. 2008; Niinemets and Hang 2009; Walker et al. 2009).
It should be noted that a “rounded-off” number is typical for any arbitrary reference value. The chronozones
provide a good example here, with values such as 5000 or 10,000. Less round, however, are
8000, 9000, 11,000, 12,000, and 13,000 Conv BP. Even less so is 2500 Conv BP, but swift climate
changes, ending with the Last Glacial, have forced the value of 11,800 to be referred to within a precision
of hundreds of years.
“Calibrating” an arbitrarily defined age (e.g. 5000 Conv BP) is different from calibrating a typical
14C date. The chronozone boundaries have essentially no error. In a sense, it is an arbitrary definition
to give an unequivocal reference value.
For about half of Mangerud’s boundaries, it is possible to simply determine the calibrated age from
the calibration curve (second column in Table 1). The function Tconv = f(Tcal) is one-to-one in many
cases, with a total duration of 27% of calendar time over the last dozen millennia. However, not only
due to the fact that 2 boundaries (2500 and 8000 Conv BP) produce essentially many calibrated
dates, the general “calibration” should be used here, and the longer parts of the calibration curve,
particularly around the boundary, transformed to cal BC. Also, chronozones are long in duration in
comparison to the typical time span of the calibration curve wiggle. The boundary between 2 chronozones
reflects considerably slower climate changes (perhaps excluding the Preboreal/Younger
Dryas). Calibrated boundaries should not depend on a 14C concentration that is strictly local in time.
1640 A Walanus & D Nalepka
The “calibrated” age of the boundary has to reflect its original, 14C conventional counterpart in as
many aspects as possible.
Figure 1 The general idea of “calibration” of Mangerud’s boundary, exemplified by 5000BP - SB/AT one. For each (calendar)
year, the responding 14C age is taken from the calibration curve and the measurement Gaussian curve is calculated
(Figure 2) and multiplied by the Gaussian weight presented for that figure. Then, all the measured Gauss are totaled, and
the median of probability densities is calculated, for the given measurement error and central position of Gaussian weight
for the samples (here 3800 cal BC). The obtained median values are plotted in Figure 3.
Table 1 Readouts from the plots (Figure 1) for selected , the proposal of the “general” value valid
for any , and final proposal of “calibrated” values with “rounding off” applied (with one ambiguity
at 7000 BC). The value  = 0 refers to the simple transformation of the precise conventional age
according to the calibration curve, which sometimes gives a unique value (neglecting calibration
curve error) and sometimes not (then including that error). Calibration curve applied: IntCal09
(Reimer et al. 2009).
BC
Conv BP  = 0 = 20 = 50 = 100 “General”  Rounded off
2500 760–550 690 660 640 640 600
5000 3780 3780 3780 3780 3780 3800
8000 7400–6850 7000 6960 6940 6940 6950 (7000)
9000 8250 8250 8240 8220 8200 8200
10,000 9650–9420 (9450) 9470 9500 9520 9540 9500
11,000 11,120–11,000 10,965 10,970 10,980 10,980 11,000
11,800 11,740 11,740 11,730 11,715 11,720 11,700
12,000 11,880 11,890 11,900 11,900 11,900 11,900
13,000 13,570
(13,900–13,200)
13,560 13,580 13,600 13,600 13,600
Calibration of Mangerud’s Boundaries 1641
PROPOSED SOLUTION
The method of defining (calibrating) Mangerud’s boundaries, in terms of calendar age, is as follows.
Let us assume a uniform distribution of samples at the calendar time axis. (It is worth mentioning
that this is an assumption for a priori distribution of ages in the typical calibration procedure.) Since,
at the moment, we are interested in one boundary (say 5000 Conv BP), we focus on samples distributed
around the expected equivalent calibrated age, maybe 4000 or 3800 cal BC.
Let us take samples of true age equal to the consecutive years: … 4001, 4000, 3999, 3998 cal BC,
…, one sample per year. However, its weight in the following calculations will diminish, according
to the bell curve (Figure 1), with the distance from the assumed cal BC boundary (e.g. 3800 cal BC
for 5000 Conv BP). Equally, the samples should be distributed Gaussially, more densely at 3800 cal
BC, less densely at 2800 cal BC, even less so at 1800 cal BC, and so on. The choice of the bell curve
for the weight is arbitrary here, and has no connection with the probability distribution of the conventional
date. The  of the Gaussian weight is also arbitrarily chosen to be 1000 yr. It has been
checked that the final result does not depend heavily on that choice. (For the 2500 Conv BP boundary,
the weights are cut symmetrically because of the AD 1950 limit.)
Figure 2 For each calendar year, the responding 14C age is calculated and the Gaussian curve is
obtained. All Gausses are totaled with the weight depending on calendar age (see Figure 1). Finally, the
median of the sum is calculated. Only 2 groups of “samples” are presented for clarity (taken for extreme
calibration curve slopes).
1642 A Walanus & D Nalepka
Each sample is recalculated from the cal axis to the 14C axis (Figure 2). For each sample, the typical
measurement Gauss is obtained, with some measurement error . The amplitude of the probability
density function is scaled down according to the weight indicated in Figure 1. All (weighted) Gaussian
curves are totaled, and for the obtained function the median is found. That median is the final
result, at that stage. Note that the result (radiocarbon age BP) is obtained for the assumed potential
cal BC value for the given boundary (e.g. the aforementioned 3800 BC), and for the assumed date
error . Calculations are performed for 20 values of  equally (logarithmically) distributed over 10
and 300 yr. The procedure is repeated for all sensible cal BC values for a 20-yr sequence (…3800,
3820, 3840 BC…).
Figure 3 shows the results of the calculations. Namely, the median Conv BP value is plotted as a
function of the measurement , with the cal BC value representing the parameter of curves. According
to the above, one median is recalculated to another median, which seems to be correct. The purpose
of any boundary is to divide. If our aim is to recalculate Conv BP to cal BC, the samples
equally divided in the sense of Mangerud’s Conv BP should also be equally divided in exactly same
manner when their real time is considered.
RESULTS
A feature of the result given in Figure 3 is the non-negligible dependence on  (Walanus 2009). If
the curves at the plot were flat, the final answer would be straightforward. Closest to that ideal is the
Subboreal/Atlantic boundary at 5000 Conv BP, as well as that of the Older Dryas/Allerød at 11,800
Conv BP. Keeping in mind that the question of “rounding off” numbers is important here, the value
of 3800 cal BC seems to be a very good candidate for the “calibration” of 5000 Conv BP.
For 9000 Conv BP, the situation is not as promising, especially in comparison with the normal calibration
result for that value (assuming  = 0, see Figure 3). In any case, despite the unique crossing
of the calibration curve at 8246.5 cal BC, which should be “rounded off” to 8250 cal BC, the value
8200 cal BC seems to be a better solution. The “exact” value is valid only for a measurement error
of zero and for very small errors. However, the realistic errors of 14C determination (14C concentration
measurement in the sample’s proper chemical fraction) are larger (Buck et al. 2003; Scott et al.
2007).
Figure 3 is a pure result of calculations, performed under the numerical assumption of a 1000-yr
sigma for the Gaussian weight of the samples (Figure 1). However, plotted lines give no unique
answer for the “calibration” of boundaries. Table 1 shows some values taken from the plots
(Figure 3) as well as some generalizations, in the 2 far-right columns, separated from the objective
values by a vertical line. The attempt to choose 1 value for the cal BC age, which would be universally
accepted for any , is the most disputed point in the procedure. The subjective weights for different,
practical, realistic values of were applied. The idea of a “realistic” measurement error is
connected with any possible “gross” errors as well as, for example, archaeological difficulties with
the sample context. The next step, given in the last column of Table 1, i.e. that of “rounding off”
numbers, is not so doubtful.
Table 2 shows the final proposal for the “calibration” of Mangerud’s chronozones. There is only one
difference in relation to the last column of Table 1, i.e. the subjective, deeper “rounding off” of the
Boreal/Atlantic border.
The ambiguity of the proposed values is clearly visible in Figure 3. The range defined in the columns
“ = 20” and “= 100” (Table 1) can be treated as error estimation for the given BC boundary.
Calibration of Mangerud’s Boundaries 1643
Figure 3 The 14C (median) age as a function of the measurement error , for different calendar (BC) ages of the trial
“center” of the large group of samples. Because lines are not strictly horizontal, the “calibration” of Mangerud’s boundaries,
in some cases only, is almost independent of . (The “conventional” description of the vertical axis is somehow
incorrect here due to the different meanings of the word “determination.”) Standard calibration (OxCal 4.1) of discussed
boundaries, obtained with  = 0, is given for comparison purposes.
1644 A Walanus & D Nalepka
A confidence interval should be attached to any given range, with the confidence level associated
roughly with the frequency of dates of  > 100 yr, in a given project. However, it should be repeated
that the idea of error or any imprecision is inconsistent with the idea of an arbitrary value. It would
only be the question of an arbitrary decision as to which strict “calibrated” numbers are to replace
the conventional 14C ages of the Holocene and Younger Dryas boundaries.
ACKNOWLEDGMENT
This work was financed by Statutory Research.
REFERENCES
Table 2 Final proposal for the “calibration” of Mangerud’s chronozones.
Boundary Conv BP BC
Subboreal / Subatlantic 2500 600
Atlantic / Subboreal 5000 3800
Boreal / Atlantic 8000 7000
Preboreal / Boreal 9000 8200
Younger Dryas / Preboreal 10,000 9500
Allerød / Younger Dryas 11,000 11,000
Older Dryas/ Allerød 11,800 11,700
Bølling / Older Dryas 12,000 11,900
Middle Glacial / Bølling 13,000 13,600
Anderson DE, Goudie AS, Parker AG. 2007. Global Environments
through the Quaternary Exploring Environmental
Change. New York: Oxford University
Press. 359 p.
Berglund BE, Gaillard M-J, Björkman L, Persson T.
2008. Long-term changes in floristic diversity in
southern Sweden: palynological richness, vegetation
dynamics and land-use. Vegetation History and Archaeobotany
17(5):573–83.
Buck CE, Higham TFG, Lowe DJ. 2003. Bayesian tools
for tephrochronology. The Holocene 13(5):639–47.
Haynes Jr V. 2008. Younger Dryas “black mats” and the
Rancholabrean termination in North America. Proceedings
of the National Academy of Sciences USA
105(18):6520–5
Mangerud J, Andersen ST, Berglund BE, Donner J. 1974.
Quaternary stratigraphy of Norden, a proposal for terminology
and classification. Boreas 3(3):109–26.
Niinemets E, Hang T. 2009. Ostracod assemblages indicating
a low water level episode of Lake Peipsi at the
beginning of the Holocene. Estonian Journal of Earth
Sciences 58(2)133–47.
Reimer PJ, Baillie MGL, Bard E, Bayliss A, Beck JW,
Blackwell PG, Bronk Ramsey C, Buck CE, Burr GS,
Edwards RL, Friedrich M, Grootes PM, Guilderson
TP, Hajdas I, Heaton TJ, Hogg AG, Hughen KA, Kaiser
KF, Kromer B, McCormac FG, Manning SW, Reimer
RW, Richards DA, Southon JR, Talamo S, Turney
CSM, van der Plicht J, Weyhenmeyer CE. 2009.
IntCal09 and Marine09 radiocarbon age calibration
curves, 0–50,000 years cal BP. Radiocarbon 51(4):
1111–50.
RybníËek K, RybníËková E. 2008. Upper Holocene dry
land vegetation in the Moravian-Slovakian boreland
(Czech and Slovak Republics). Vegetation History
and Archaeobotany 17:701–11.
Scott EM, Cook GT, Naysmith P, Bryant C, O’Donnell D.
2007. A report on Phase 1 of the 5th International Radiocarbon
Intercomparison. Radiocarbon 49(2):409–
26.
Sørensen R. 2008. Late Weichselian deglaciation in the
Oslofjord area, south Norway. Boreas 8(2):241–6.
Walanus A. 2009. Systematic bias of radiocarbon
method. Radiocarbon 51(2):433–6.
Walker M, Johnsen S, Rasmussen SO, Popp T, Steffensen
J-P, Gibbard P, Hoek W, Lowe J, Andrews J, Björck S,
Cwynar LC, Hughen K, Kershaw P, Kromer B, Litt T,
Lowe DJ, Nakagawa T, Newnham R, Schwander J.
2009. Formal definition and dating of the GSSP (Global
Stratotype Section and Point) for the base of the
Holocene using the Greenland NGRIP ice core, and
selected auxiliary records. Journal of Quaternary Science
24(1):3–17.
Wanner H, Beer J, Bütikofer J, Crowley TJ, Cubasch U,
Flückiger J, Goosse H, Grosjean M, Joos F, Kaplan
JO, Küttel M, Müller SA, Prentice IC, Solomina O,
Stocker TF, Tarasov P, Wagner M, Widmannm M.
2008. Mid- to Late Holocene climate change: an overview.
Quaternary Science Reviews 27(19–20):1791–
828.
Webb SD, editor. 2006. First Floridians and Last Mastodons:
The Page-Ladson Site in the Aucilla River.
Topics in Geobiology 26. Dordrecht: Springer. 588 p.