Professional Documents
Culture Documents
apilarinos,+207+Times+Up+searchable Part23
apilarinos,+207+Times+Up+searchable Part23
Sturt W Manning
Khania 15/TR10,Rm E charred seed Pisum sativum OxA-2517 3380 80 LMIB Early to
Late?
Khania 13/TR17,1984,Rm C charred seed Viciafaba OxA-2518 3340 80 LMIB Early to
Late?
Khania 14/TR17,1984,Rm C charred seed Hordeum sp. OxA-2646 3315 70 LMIB Late?
Khania 16/TR24, 1989,L6,BA1 charred seed OxA-2647 3315 70 LMIB Late?
Khania 13/TR17,1984,Rm C charred seed Viciafaba OxA-10320 3208 26 LMIB Early to
Late?
Khania 14/TR17,1984,Rm C charred seed Hordeum sp. OxA-10321 3268 27 LMIB Late?
Khania 15/TR10,Rm E charred seed Pisum sativum OxA-10322 3338 26 LMIB Early to
Late?
Khania 16/TR24, 1989,L6,BA1 charred seed OxA-10323 3253 25 LMIB Late?
Myrtos-Pyrgos 17 /K5,2, 1 charred seed Hordeum sp. OxA-3187 3230 70 LMIB Late
Myrtos-Pyrgos 18/K5,2,4 charred seed Hordeum sp. OxA-3188 3200 70 LMIB Late
Myrtos-Pyrgos 19/K5/K6,2,1 charred seed Vicia ervilia OxA-3189 3270 70 LMIB Late
Myrtos-Pyrgos 20/K5/L6,2,2 charred seed Vicia ervilia OxA-3225 3160 80 LMIB Late
Myrtos-Pyrgos 17 /K5,2, 1 charred seed Hordeum sp. OxA-10324 3270 26 LMIB Late
Myrtos-Pyrgos 19/K5/K6,2,1 charred seed Vicia ervilia OxA-10325 3228 26 LMIB Late
Myrtos-Pyrgos 20/K5/L6,2,2 charred seed Vicia ervilia OxA-10326 3227 25 LMIB Late
Myrtos-Pyrgos 18/K5,2,4 charred seed Hordeum sp. OxA-10411 3150 40 LMIB Late
Knossos MUM charred seed Hordeum sp. OxA-2096 3070 70 LMII Advanced
Knossos MUM charred seed Hordeum sp. OxA-2097 3190 65 LMII Advanced
Knossos MUM charred seed Hordeum sp. OxA-2098 3220 65 LMII Advanced
Knossos MUM charred seed Hordeum sp. OxA-11882 3156 33 LMII Advanced
Knossos MUM charred seed Hordeum sp. OxA-11943 3148 23 LMII Advanced
Table 1. Radiocarbon Dates on (i) short-lived samples fi·om Late Minoan IB contexts at Khania, Mochlos and
Myrtos-Pyrgos (after Housley et al. 1999; Bronk Ramsey et al. 2004; Manning et al. 2006a; Soles 2004b); and (ii)
radiocarbon dates on short-lived samples from the Advanced LM II destruction at the Minoan Unexplored Mansion
at Knossos (after Housley et al. 1990, 21 4-215; Hedges et al. 1990, 227; Bronk Ramsey et al. 2004; Manning et al.
2006a, b). All are AMS dates except sample 115890 from Mochlos listed as radiometric (i.e. routine 14 C dating);
Mochlos samples 85991 and 85992 were run at Oxford after pretreatment at Beta Analytic (Soles 2004b, table 40),
the others at Beta Analytic itself or at other unnamed laboratories following pretreatment at Beta Analytic (Soles
2004b, 145). All the other (OxA) samples were pretreated and run at the Oxford Radiocarbon Accelerator Unit.
The Oxford samples come from a lab with published pretreatment regime and with published known age test results
indicating good accuracy and precision (information relevant to the Khania and Knossos and Myrtos-Pyrgos samples
can be found in the Manning et al. (2006a) paper- see Supporting Online Material, Manning et al. 2006b). We do
not have the same level of information for the pretreatment procedures for the Beta dates, nor for the un-named other
laboratories/ accelerators. We do not have published information on the performance quality of the Beta radiometric
dates (re sample 115890).
J},
"' 'Y Stt frt 1111111111~
~ ~I'v~ ruu:;-Pyr ~u:; and vfochlos pnly
~·
n ,-J,"Y ~MIBu ~e •Lk TB Fiml_
Fig. 1. Myrtos-Pyrgos and Mochlos, LM IB Late and LM IB Final, sequence analysis in isolation. OxCal (Bronk
Ramsey 1995; 2001; 2008a) with IntCal04 (Reimer et al. 2004). The hollow distributions for each individual date
show the calibrated calendar age probabilities in isolation (no model), and the solid distributions show the reduced
p robability distributions after applying the sequence analysis model shown. T he upp er and lower lines under each
distribution show the respective la (68 .2 % confidence) and 2a (95.4% confidence) calibrated age ranges (for the
modeled results): see text for details. The agreement index compares the final (posterior) distribution calculated (the
solid histogram) against the original distribution (the calibrated age probability for the individual sample in isolation:
the hollow histogram). If the former is unaltered the index value is 100. T he value rises above 100 where the final
distribution overlaps only with the very highest part of the prior distribution . In contrast, an agreement index below
60 indicates disagreement with the model (and insufficient overlap of the distributions) at abo ut the 5% level of a chi-
squared test. The overall agreement index for each sequence is also stated - again a score greater than the stated test
statistic indicates that the m odel surpasses an approxim ate 95% confidence leveL See p ostscript.
cally prior to the Mochlos data. T his archaeological gos and Mochlos each seem to represent the same
sequence information can thus be employed in a time horizon at their respective sites at 95% confi-
Bayesian radiocarbon analysis to gain greater reso- dence (thus same year or year two of growth for the
lution and precision for the dating of the LM IB seeds in question) : yielding weighted averages of (i)
p eriod. This paper employs the OxCal software 15
and the IntCal04 16 and lntCal98 17 radiocarbon 15
Bronk Ramsey 1995; 2001; 2008a. Plots and data from version
calibration datasets (with curve resolution set at 5 3.10, current when this paper was first written. See Postscript.
and rounding - to nearest 5 years - " on"). Fur- 16
Reimer et al. 2004.
ther, the sets of short-lived data from Myrtos-Pyr- 17
Stuiver et al. 1998 .
R_Cor foinP m
J~
· 1 HU nu.:> (L vliBFinc. 1
/) 965~!:Q_ r~ --=.,
/UMUOOW
-/, TMTRIT l;fJJ ,, w •ouv, l..al.. I
I
D.
~
"o'~';~E 7d ~-JI
I I
2100BC 2CXXlBC 1900BC 1800BC 1700BC lro:JBC 1500BC 1400BC 1300BC
Calendar Date
Myrtos-Pyrgos (n=8) of3229±13 BP with a x2 test LMIB Late destruction is placed 1525-1490 BC (1a)
statistic of7.6 < 14.1 the 95% confidence value for and the Mochlos destruction (that is a Late Minoan
7 degrees of fi·eedom, and (ii) Mochlos (n=5) of IB Final destruction) is placed 1480-1440 BC (1 a)
3220±20 BP with a x2 test statistic of 3.4 < 9.5 the (2a, 95.4% confidence: 1530-1455 BC and 1505-
95% confidence value for 4 degrees of freedom. 18 1430 BC respectively).
This is nicely consistent with the idea that these The whole/nujority of Late Minoan IB Late
short-lived samples come from (i.e. were harvested is therefore bifore c.1525-1490 BC (or at 2a before
and stored/used shortly before) the respective de- 1530-1460/55 BC). And all of Late Minoan IB
struction contexts. The weighted average of each Early lies bifore this . How long is the LM IB Final
set of data thus offers the best estimate for the rel- phase? How long is the LM IB Late phase? How
evant year(s) of growth for the samples from each long is LM IB Early? We do not know (and, note to
destruction level. excavators: we urgently need Late Minoan IB Early
Let us begin by considering just these two phased radiocarbon data). But it seems highly unlikely that
data sets: Figs 1 and 2. We find that the Myrtos- these phases are to be measured in terms ofless than
Pyrgos destruction set (that is a Late Minoan IB a few decades each, and one or more could well
Late destruction) is placed (i) fi·om IntCa104: 1525- represent several decades to a half-century of time.
1490 BC (1a, 68.2% confidence) and the Mochlos Rutter (forth.) suggests at least a couple of decades
destruction (that is a Late Minoan IB Final destruc- but probably no longer than 50 years for either of
tion) is placed 1485-1445 BC (1a) (2a, 95.4% con- each LM IB Late and LM IB Final, so maybe 50-
fidence: 1530-1460 BC and 1500-1430 BC respec-
tively); or (ii) from IntCal98: the Myrtos-Pyrgos IH Ward & Wilson 1978.
Prase
f-+- -+-
-+- -+-
G~-2517 90.1 %
~
c'KA-2518 107. ~% ~l
·-+- I
f-+- -t-
&A-2646 110. , %
~~
G'rA.-2647 87. ( % ~
-+-
c'KA-10320 90. 7% J~
-=--
c -A-10323 10 1% JJL
-+- T
E ent Khania --- .4~
Bound. ,7Y _Bound __j~
.=_____..
.
I
30000C 25000C 20000C 15000C lOOOOC
Calendar Date
Fig. 3. A Phase (a gro up of data for which we have no information about their respective relative ages vis a vis each
other, but which we can define as a grouping vis avis other information - in this case the data all come from LM IB
destruction contexts at Khania and are assumed to form a time horizon - LM IB - at Khania) analysis of the eight
radiocarbon data from LM IB contexts at Khania (employing IntCal04 and OxCal) . An Event (see OxCal manual:
http :/ / c14.arch.ox.ac. uk/ oxcalhelp/ hlp_contents.html) describing the Phase comprising the eight radiocarbon dates
fi:om Khania LM IB destruction contexts is shown - this attempts to define the data within the phase (inside the
boundaries). The date ranges calculated for the Event are cited in the text. The spread reflects the range of older
through later ages within the Khania set evident in the eight individual calibrated age ranges shown above. For general
description of how to read the plot, see caption to Fig. 1.
a minimum exclusion criterion to yield a set with a I consider Sequence analyses below with three
weighted average which satisfi es a C hi-squared test options: (i) Khania treated as one overall set and
at the 95% confidence level, 24 then excluding just thus as one OxCal Phase (n= S) = Model 1, (ii)
OxA- 10320 allows the other seven dates to yield Khania treated as the minimum coh erent set of
a weighted average of 3293±14 BP (T=8.2 < 12.6 seven data with a weighted average of3293 ±14 BP
for df6). This would place the Khania set as rather = Model 2, and (iii) Khania treated as two groups,
older than the M yrtos-Pyrgos LM IB Late destruc-
tion or M ochlos LM IB Final destruction sets. 24
Ward & Wilson 1978.
~·'··
CM-2517 101.7% description ofhow
to read the plot, see
CM-2 18 lll.IY/o
caption to Fig. 1.
CM-26,46 112.1%
CM-2647 761%
CM-JCIJ20 94.2%
CM-JCIJ21 102.3%
CM-1q322 965%
CM-10323 105.5%
Event Nhama IMB
1. Pl:Ja&~os-PyrwsO~late)
R_Conlpinelvfrtos-Pyt~os (IMBLate) 103.3%-~~-
Boundaiy IJJp3 Late to IM1 Finnl ~
an "Older" group (O:x:A-2517 , 10322) , and a "Lat- Khania destruction data (or Khania "Older" >
er" group (O:x:A-2646, 10321 , O:x:A-2647 , 10323) Khania "Later") ~ Myrtos-Pyrgos destruction data
(n=6 and excluding the somewhat divergent ages > Mochlos destruction data > Late Minoan 11 de-
on th e sample determined by O :x:A- 2518 and struction data.
10320) = Model3. This sequence comprises samples from four dif-
Finally, as an important constraint on the latest ferent sites from all over Crete. No special circum-
possible placement of the LM IB data, we have a set stances apply (like claims of possible volcanic C0 2
of data on barley samples from the Advanced LM effects, etc., unusual wiggles/plateau in the radio-
11 destruction of the Minoan Unexplored Mansion carbon calibration curve, etc.).
(MUM) at Knossos (Pop ham 1984): see Table 1. We may use this archaeological sequence to in-
This context is (by an unknown amount) later than form a Bayesian analysis of the likely calendar cali-
all the LM IB contexts. brated age ranges for the data. Figs 4-9 show the
We therefore have a Late Minoan IB to Late Mi- calibrated age range analyses for this overall LM
noan 11 sequence of: IB-11 sequence (given the three options for treating
~
I I I I I
Fig. 5. As Fig. S:x}..ence
r 1
{A=100.3o/<(~c=60. I ~}
I
4 but using
IntCal98. Boundmy Stm1 .........:. I
S:x}..ence 11\ ~II
P~IJ\ ~Early to l.at 0-ete
Phm:e <hania (IMBI farly? to late)
1- 1-
CM-2 17103.3% ~
1- 1--
CM-2~I8 I12.00/o ~
+-- 1-
- I
-~
CM-2(546 111.2% ~
\- -
CM-2 ~7 83.6% i~
~
t - 1-
1-
_,_ -
CM-1 tmo 99.1%
CM-1 'D321 102.5%
CM-I 0322 96.4%
CM-I 0323 101.5%
.. ...._
li-\
JtJ....._______
~~
I
1-
~
Event ~1aniaiMB
\-
P~ M'!-tos-Pyrws \IMBI.ate) I
~-[
_].
-
J R eo,bbine1vpl1os-Pj,gos(IMBlitt. P 100.~ I
BmmdcnyL (;m Inte to IM SFJnal I
[ P~M~hlos(IMBFi lal) I
1-
' -I R_Combine Mxhlos (IJv, rJJFJnap 91.00/o l!i I
I
[ P~la eMnoan II
--
j
R Combipe Knossos !MJ. IRstruction sha 1-lived 1fl? m I
a;;:
BmmdmyEnd
t
3500IC 3000IC 2500IC 2000IC 1500IC lOOOIC
Chlencbr Date
Khania described above, and considering both the Akrotiri on Thera - but, since we know archaeo-
IntCal04 and IntCal98 calibration datasets). The logically that they must be later, this implies that
date ranges calculated are shown in Table 2. the mid-16'" century BC "w iggle" that creates a de-
We see fi:om Figs 4-9 and Table 2 that the ra- gree of am.biguity in the dating of the Santorini/
diocarbon data form a good analysis with the clear Thera evidence is perhaps the cause of these LM IB
archaeological Myrtos-Pyrgos > Mochlos > Knos- dates which seem rather similar in radiocarbon age
sos sequence consistent w ith the radiocarbon evi- to those from late in the LM IA period. If so, we
dence (good agreement index values). The Khania might speculate that the Khania "Older" dates lie
data are less constrained, but are also compatible, on the wiggle in the radiocarbon calibration curve
and provide further evidence for some parts of the c. 1575-1535 BC. 25 These data from Khania LMIB
LM IB period likely lying through much if not all destruction contexts, although labeled "Older"
the 16'" century BC. (The Khania LM IB "Older" here -- versus the other Khania dates - do not seem_
dates offer radiocarbon ages contemporary with
those fi·om the LM IA volcanic destruction level at 25
See previously Manning 1992; Housley et al. 1999.
to come from the LM IB Early phase defined by LM 11 older. This implies, even if the LM 11 period
Rutter - they were submitted as final LM IB de- is considered relatively short, an earlier to mid-15th
struction. We might assume that at least the LM IB century BC start date for LM 11 (and the end ofLM
.• Early Phase, and maybe even LM IB Developed, lie IB).
bifore the Khania "Older" set -likely in the first half T he dates of the Mochlos LM IB Final destruc-
of the 16th century BC.) tion, and the Myrtos-Pyrgos LM IB Late destruc-
The terminus ante quem for the LM IB period tion, contexts are very consistent across all the sce-
from the Advanced LM 11 destruction at the MUM narios in Figs 1-2 and 4-9. Taking the 1a ranges
at Knossos is clear and specific taking the most like- (or main range therein) as the most likely indica-
ly 1a ranges from Table 2: between c. 1440/1435 to tive reality: the very end of LM IB Final (Mach-
1411 / 05 BC. Moreover, we must also allow for the los) lies between c.1500/ 1485/81/80/75/70 to
fact that much (or most) of the LMII period lies be- 1455/ 49 / 45/44/40/35 BC. The whole LM IB pe-
fore the (Advanced LM 11) destruction event dated riod (that is each of the Early, "Developed", Late,
by these samples -making the effective likely termi- and Final phases) lies before this. The destruction
nus ante quem for the end of LM IB or the start of context ofLM IB Late at Myrtos-Pyrgos is various-
I ,n1:Jermt.hPrmt--1l>n5.J..!.,J%~b,M""~J--/·-
R_Can1 'li1E Kfvn 'a IMRVl&p
Pliffielate ~II
/.fuwdar End
l
2200BC 2100BC 2<mBC l<xXJBC 1800BC 1700BC 16XJBC 1500BC 1400BC 1300BC 1200BC
Ollrndar Date
ly placed c. 1525/21/20/19/15 to 1498/95/90/85 One could therefore be more conservative and cite
BC. Again, most of LM IB Late (before this close of just the 2a ranges- the most likely 95.4% of the to-
phase destruction), LM IB "Developed", and LM tal probability. These numbers are given in Table 2
IB Early lies beforehand. or in the text above. Thus the Mochlos LM IB Final
I note that the above are date ranges encompassing destruction dates c. 1515/10/08/05/04/1500/1490
the most likely 68.2% of a 100% probability. It is not to 1440/38/35/30/25 BC and the Myrtos-Pyrgos
legitimate to glance at them and then to choose to LM IB Late destruction dates c. 1530/29/27/25 to
pretend that the last year of the range is a reasonable 1466/65/61/60/55 BC. In each case, these 2a rang-
number to use. Indeed, years more within the range es widen the overall range both up and down. They
are more likely (depending on the exact shape of the do not change the clear indication to be drawn from
probability distribution: see these - the solid histo- these data. Seeking to cite the very end of the 2a
grams- in Figs 1-2, and 4-9). One must consider the range and ignoring the rest of the range - and espe-
range. In the previous two paragraphs I cited the 1a cially the most likely 68.2% part- is misleading Qust
ranges. These are the most likely 68.2% of the dating as if one cited just the very top end of any of these
probability. But of course there is the other 31.8%. ranges).
'I/11 ·bi}Start
~ IPnceTMP ~TT
...... .....
with IntCal04. The
Model3 LM IB-II
sequence treats the
Khania data as two
separate groups,
Khania "Older" and
IPhare Khani ilMB'Ear ~y' Khania "Later" (see
text for discussion) .
R_ Combine r<hania 'Fan "' 07 )0/n
·!~ For general
P~Late~ Ainmnii
1 Q8. )0/n
••
R_Combine Z.
1IM 7 IRstructio1 rshort-!iiP;{ 7())7~
·--=
Buurlf. rwFnd
-4;~
24000C 22000C 20000C 18000C 16000C 14000C 12000C
Chlenchr Date
All the data in Table 2 indicate a very similar IB-11 case, all the evidence points to IntCal04 be-
message, and there is only a little difference between ing the best most appropriate radiocarbon calibra-
using IntCal04 or IntCal98. IntCal04 includes ad- tion dataset presently available (there will of course
ditional data and was constructed with a rigorous be further revisions to the international radiocar-
statistical procedure compared to the ad hoc ap- bon calibration curve in the future). From the ar-
proach employed for lntCal98. 26 The approach in chaeological perspective, there are Model 1, Model
lntCal04 slightly smoothes some of the "ragged" 2 and Model 3 in Table 2. Models 2 or 3 provide
nature of lntCal98 - as a result occasionally it may narrower dating estimates for the Khania contexts.
lose a little sensitivity for tree-ring radiocarbon
wiggle-matching exercises . But for general pur- 26
See Reimer et al. 2004; Buck & Blackwell 2004; Blackwell
poses, and for archaeological situations like our LM et al. 2006.
Sxpnce TlMrR-rr
P~ l lmalM B'Early'
'[c,ntlUf'"~-----:::::~~~-j--
Boundmy r<hania I.lvJVJ 'Early' tc IMBLL
p~ l\Yhchlos(UVIDFmal.)
P~ I late Mnoar II
Of these, Model 2 offers the wider and somewhat destruction evidence. This date range, and as in-
earlier overall range, but its peak probability in the dicated by at least the later group of radiocarbon
nwdeled analysis (Fig. 6) lies c. 1531/1530 BC, not dates, and the ceramics from the LM IB destruc-
really that far from the 1525-1500 BC range for the tion at Khania, is assun'led to be relevant roughly to
"Late" Khania subset employed in Model3 . On the some part of the LM IB Late phase in terms of the
basis that choosing as late a date as possible for the Rutter scheme.
Khania context is appropriate given this involves Thus, for a best (current, working) rounded
then no favouritism towards a "High" chronology approximation of the dates of our contexts, we
(and instead deliberately favours a minimum chro- might cite an amalgamation of the Model 2 and 3
nology), we might estimate a date of c. 1530-1500 IntCal04 1o results, thus:
BC for at least the later part of the Khania LM IB
Model 3 - Khania treated as Khania "Early" and Khania "Late" (see text) (Figs 8-9)
Site-context la IntCal04 BC 2a IntCal04 BC la IntCal98 BC 2a IntCal98 BC
Chronology. 78
Conclusion Acknowledgements
The evidence for a long multi-phase LM IB period I thank Jeremy Rutter for information and
on Crete (Rutter forth.), and the evidence of the comments, the editor David Warburton, and Erik
LM IB radiocarbon dates (above), clearly under- Hallager.
mine the Low Aegean Chronology. A start for the
period no later than the early to mid 16'h century
BC seems necessary (ignoring any other evidence).
A start at the end of the 17'11 century/start 16'h cen- 78
Manning et al. 2006a; 2009; Friedrich et al. 2006; 2009; and
papers in this volume with W Friedrich as an author.
tury BC would seem entirely reasonable £i·om the 79
Manning et al. 2006a; 2009; Friedrich et al. 2006; 2009; and
evidence summarized in Table 2 (remembering the papers in this volume with W Friedrich as an author.
dates there are for LM IB Late and LM IB Final de- 80
Manning et al. 2006a; and with LM IB-II as further elu-
structions and therefore that much or all of the over- cidated in the present paper. Note: there is some definition-
all LM IB period lies bifore these date ranges). The al variation over whether what was termed LM IA early at
Kommos should instead be referred to as MM IIIB (Girella
need to re-think the LM II period appears entirely
2007). If so, this "new" MM IIIB might run down to around
in accord with - and indeed more compatible with 1685-1680 BC give or take (see Manning & Bronk Ramsey
- the archaeological evidence linking the reign of this volume).
Postscript
Each run of an OxCal Sequence analysis produces especially, can vary, and also when compared
slightly different outcomes. Data in the paper to the outputs £i·om the subsequent OxCal 4,
represent average or typical values £i·om several where some minor differences in the revised
runs £i·om OxCal 3.10. The main likely ranges software also affect exact outcomes. For example,
remain fairly stable across different runs, but the considering and comparing Fig. 1 above, if OxCal
break-points where there are possible sub-ranges, 4 .11 is employed (with IntCal04) now as proofs
226