Living at the margin of the retreating Fennoscandian Ice Sheet: The early Mesolithic sites at Aareavaara, northernmost Sweden

Abstract

During an archaeological survey in Pajala parish, northernmost Sweden, clusters of quartz waste from knapping and burnt bone were discovered on a glaciofluvial gravel plateau close to Aareavaara village in the Muonio River valley. Sampled materials from a larger area and small-scale excavations (in total 6 m2) are interpreted as resulting from short-stay hunter-gatherer camps. Radiocarbon dating on burnt bones suggest an age of occupancy at ~10,700 cal. yr BP, which is more or less contemporary with ‘Komsa Phase’ sites on the north coast of Norway (~300–360 km northwards). The Aareavaara site should thus be the oldest known archaeological site to date in northern Sweden. A palaeoenvironmental reconstruction, based on pollen analysis of sediment cores from two nearby lakes and radiocarbon dating of macrofossils for construction of time/depth sedimentation curves, suggests a deglaciation age of the area corresponding to occupation by early man (~10,700 cal. yr BP). Aareavaara was at the time of deglaciation situated in a transitional zone between subaqueous and subaerial ice-margin retreat from the northeast towards the southwest, with higher hills and plateaux forming an archipelago in the Ancylus Lake with highest shorelines formed at ~170 m a.s.l. The hunter-gatherer camp sites at Aareavaara were thus, both in time and space, located in close proximity to the retreating ice sheet margin, but also in a waterfront location, in fact on an island in the Ancylus Lake. Our pollen data suggest a subarctic birch woodland tundra landscape characterized by open vegetation, including occasional birch trees and an abundance of willow and dwarf birch.

Full text

The Holocene
23(1) 104 –116
© The Author(s) 2012
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0959683612455546
hol.sagepub.com
Introduction
During the last glacial maximum (LGM, c. 26,500–20,000 cal. yr
BP; Clark et al., 2009) the Fennoscandian Ice Sheet (FIS) merged
with the Barents Sea Ice Sheet, forming the western part of the
so-called Eurasian Ice Sheet complex over areas that today are
northernmost Norway, Sweden, Finland and northwestern Russia
(Svendsen et al., 2004). This merging and the maximum expan-
sion of the FIS in its northern sector seem to have been at around
18,000 cal. yr BP (Larsen et al., 2006). After the break-up of the
Eurasian Ice Sheet the retreat of the FIS margin was subaqueous
because of isostatic depression of the crust, despite low world sea
level, giving rise to a marine margin towards the Barents Sea sec-
tor and a glaciolacustrine setting (the Ancylus Lake) in the Gulf of
Bothnia between Sweden and Finland (Figure 1). Humans could
move into freshly deglaciated areas only when these were above
the marine limit or highest shoreline. This means that such early
coastal sites/settlements usually are found far inside present
shorelines or at high altitudes in present coastal areas because of
isostatic rebound and, as a consequence, shoreline regression. To
date, the oldest known early Mesolithic sites in northernmost Fen-
noscandia are from the north and northwest coast of Norway (the
‘Komsa Phase’, e.g. Woodman, 1993). The original references
give ages in 14C years, which are here recalculated as calendar
years (cal. yr BP) according to the Oxcal v4.1 package (Bronk
Ramsey, 1995, 2001), here given as median ages of the calibrated
age range (1 σ). Selected sites shown in Figure 1 are: 1: Lagesidribakti
site at Varangerfjorden (11,400 cal. yr BP; Blankholm, 2008;
Grydeland, 2002); 4: Slettnes site (10,949 cal. yr BP; Blankholm,
2008; Hesjedal et al., 1996); and 6: Evjen 3 site at Saltstraumen
(10,930 cal. yr BP; Blankholm, 2008; Hauglid 1993). This oldest
northern colonization seems thus to have taken place during the
Preboreal, a much warmer period following the cold Younger
Dryas stadial. Mesolithic sites further inland in northern Finland
and Sweden have so far been dated to substantially younger ages,
often in the order of thousands of years younger. An exception to
this is the Sujala site in northernmost Finland (site 2 in Figure 1),
by Rankama and Kankaanpää (2008) dated to ~10,400 cal. yr BP.
A large open-air mine is in the process of being established at
Tapuli, ~20 km north of Pajala village in Norrbotten county, Swe-
den (Figure 1). In connection to this the Norrbottens museum (the
county museum of Norrbotten) was assigned by the mining com-
pany (Northland Resources Inc.) to carry out an archaeological
survey of areas that possibly could be affected in various ways by
this enterprise. The principal survey was carried out in 2009, at
455546HOL23110.1177/095968
3612455546The HoloceneMöller et al.
2012
1Lund University, Sweden
2County Museum of Norrbotten, Sweden
3University of Copenhagen, Denmark
Corresponding author:
Per Möller, Department of Geology, Lund University, Sölvegatan 12, SE
223 62 Lund, Sweden.
Email: per.moller@geol.lu.se
Living at the margin of the retreating
Fennoscandian Ice Sheet: The early
Mesolithic sites at Aareavaara,
northernmost Sweden
Per Möller,1 Olof Östlund,2 Lena Barnekow,1 Per Sandgren,1
Frida Palmbo2 and Eske Willerslev3
Abstract
During an archaeological survey in Pajala parish, northernmost Sweden, clusters of quartz waste from knapping and burnt bone were discovered on a
glaciofluvial gravel plateau close to Aareavaara village in the Muonio River valley. Sampled materials from a larger area and small-scale excavations (in total
6 m2) are interpreted as resulting from short-stay hunter-gatherer camps. Radiocarbon dating on burnt bones suggest an age of occupancy at ~10,700
cal. yr BP, which is more or less contemporary with ‘Komsa Phase’ sites on the north coast of Norway (~300–360 km northwards). The Aareavaara
site should thus be the oldest known archaeological site to date in northern Sweden. A palaeoenvironmental reconstruction, based on pollen analysis of
sediment cores from two nearby lakes and radiocarbon dating of macrofossils for construction of time/depth sedimentation curves, suggests a deglaciation
age of the area corresponding to occupation by early man (~10,700 cal. yr BP). Aareavaara was at the time of deglaciation situated in a transitional zone
between subaqueous and subaerial ice-margin retreat from the northeast towards the southwest, with higher hills and plateaux forming an archipelago in
the Ancylus Lake with highest shorelines formed at ~170 m a.s.l. The hunter-gatherer camp sites at Aareavaara were thus, both in time and space, located
in close proximity to the retreating ice sheet margin, but also in a waterfront location, in fact on an island in the Ancylus Lake. Our pollen data suggest a
subarctic birch woodland tundra landscape characterized by open vegetation, including occasional birch trees and an abundance of willow and dwarf birch.
Keywords
Ancylus lake, deglaciation, early Mesolithic, Fennoscandian ice sheet, hunter-gatherer, palaeoenvironment
Received 24 January 2012; revised manuscript accepted 28 May 2012
Research paper

Möller et al. 105
which a number of archaeological sites were identified. Special
interest was drawn to two sites (Raä 1276 and Raä 1277 in Pajala
parish) close to the small village of Aareavaara, lying on the
southern bank of the Muonio River, marking the border with Fin-
land (Figure 2). Here, on a flat sand and gravel plateau rising
above surrounding wetlands, quartz waste and burnt bone were
exposed in forest plough marks, suggesting human occupation.
Two preliminary 14C datings on found bone fragments turned out
at surprisingly old ages, 12,672–11,216 (median 11,935) and
11,326–9912 (median 10,733) cal. yr BP, respectively (Table 1).
At face value such old ages seemed to be too old to be true.
According to the ‘conventional’ deglaciation chronology of north-
ern Sweden (Lundqvist, 2009) the area should have been covered
by the Late Weichselian ice sheet at this time! This chronology
has, however, been challenged by Lindén et al. (2006), suggesting
a substantially earlier deglaciation of the Bothnian coastal areas.
And if the bone ages are correct, could it really have been possible
that humans visited the area at that time? Uncertainty about the
very old ages made archaeological excavations and geological
investigations necessary to clarify the chronology of the Aareav-
aara sites and their relation to FIS deglaciation, shoreline displace-
ment and palaeoenvironmental evolution.
Study area
The excavated sites are situated at Aareavaara, ~25 km north of
Pajala in northeastern Sweden (Figures 1 and 2). The surrounding
area is topographically a flat to slightly undulating bedrock plain,
in turn predominantly covered by flat till deposits (ground
moraine) and in depressions extensive mires, all at altitudes
between 160 and 220 m a.s.l. Dispersed bedrock hills, both north
and south of the Muonio River, reach altitudes of 250–300 m a.s.l.
In connection to the deglaciation of the area, meltwater from the
retreating ice formed flat sand and gravel plateaux east of Aareav-
aara village (Fromm, 1965), locally named Koskenkangas, on
which the archaeological remains were discovered.
Figure 1. Compilation of ice recessional lines for the Fennoscandian Ice Sheet (FIS): the Younger Dryas margin according to Mangerud et al.
(2011), recessional lines in Finland according to Johansson et al. (2011) and in Sweden according to Lundqvist (2009). The highest shoreline
and areal coverage by the Ancylus Lake at deglaciation is according to Tikkanen and Oksanen (2002) (Finland) and Lundqvist (2009) (Sweden).
The ice recessional line through Boden is marked as 10,500 cal. yr BP (according to Lindén et al., 2006, which is 500 years older than the
age for this recessional line as given in Lundqvist, 2009). Marked early Mesolithic in Fennoscandia sites > 9500 cal. yr BP are: 1: Lagesidbakt,
11,818–11,193 (median 11,400) (Blankholm, 2008; Grydeland, 2002); 2: Sujala, 10,586–10,253 (median 10,441) (Rankama and Kankaanpää,
2008); 3: Inari, 10,189–9609 (median 9910) (Halinen, 2005); 4: Slettnes, 11,193–10,724 (median 10,949) (Blankholm, 2008; Hesjedal et al, 1996);
5: Simavik, 11,089–9785 (median 10,409) (Blankholm, 2008); 6: Målsnes, 11,165–10,524 (median 10,825) (Blankholm, 2008); 7: Evjen 3, 11,195–
10,673 (median 10,930) (Blankholm, 2008; Hauglid, 1993); 8: Mohalsen, 11,325–9776 (median 10,623) (Bjerk, 1989; Blankholm, 2008); 9: Kangos,
10,112–9540 (median 9698) (Östlund, 2004); 10: Dumpokjauratj, 9890–9481 (median 9623) (Bergman et al, 2004); 11: Soumusalmi, 10,379–9632
(median 10,030) (Matiskainen, 1996); 12: Hyrynsalmi, 10,398–9672 (median 10,041) (Korteniemi and Suominen, 1998). Marked early Mesolithic
sites in NW Russia are from Dulokhanov and Khotinskiy (1984); these are not specifically named and are only given an age frame, recalculated
to calendar years to >~9000 cal. yr BP.

106 The Holocene 23(1)
At deglaciation the ice margin of this area was preferentially
subaqueous, standing in more-or-less deep water. The highest
shoreline (HS) is situated at ~170 m a.s.l. here (Lundqvist, 2009),
or slightly lower, gradually evolving as the deglacial water body
in the Bothnian Basin – a freshwater stage of the Baltic named the
Ancylus Lake (Björck, 1995) – followed the receding ice margin
until it eventually reached areas in excess of this altitude. When
the HS formed in the area the water was quite shallow; thus gla-
ciofluvial deposits were built up more or less to the contemporary
water level and the area formed an archipelago in Ancylus Lake
with more highly situated till plateaux and bedrock hills protrud-
ing above the water surface. Land uplift rate was high in connec-
tion to the deglaciation; in the Boden area, ~200 km towards
the southeast (Figure 1), it has been shown to have been ~9 m/100
yr for the first centuries after deglaciation, after which there was
an exponential reduction to today’s values of ~0.8 m/100 yr
(Linden et al., 2006). This suggests that there was a rapid shore
regression in the Aareavaara area and water-covered areas at
deglaciation quickly became dry land.
Materials and methods
Archaeological excavation and sampling
The Koskenkangas plateau (~163–170 m a.s.l.; Figure 2) was
deforested in 2001 and 2003, after which there was soil preparation
before pine replanting, i.e. shallow ploughing of the ground leaving
~1 m wide plough marks with 1.5–3 m wide zones of undisturbed
soil between (Figure 3c). The archaeological sites Raä 1276 and
Raä 1277 (Figure 3a and b), discovered as a result of exposed
quartz waste and burnt bone fragments in these plough marks, were
excavated in 2010. Four test pits covering a total area of 6 m2 were
laid out, one of them at site Raä 1276 and three of them at site 1277.
The exact positions of the test pits and finds outside these were
measured with a GPS, using the Swedish RTK (Real-Time Kine-
matics) reference network, giving a position accuracy of ±1–2 cm.
After removal of any surface turf, each test pit was excavated
with trowel in 5 cm deep layers until no man-produced artefacts
could be detected. Artefacts were marked when found (Figure 3d)
and their position manually measured (n, e, z; Sweref 99 TM)
within the test squares with carpenters’ ruler and levelling instru-
ments, measured from fixed points and coordinate sticks that
previously had been established from the GPS measurements.
All objects were registered and sampled, after which soil from
each excavation layer was sieved through a 3 mm mesh. When
finds were made in the sieve, they were measured to 0.25 m2 from
where the soil had come and examined for smaller objects (e.g.
the smallest fragments of fish bones, etc.)
Geological field work
A lake coring programme was conducted in March 2010, per-
formed from winter ice, aiming at dating the deglaciation of the
Figure 2. Relief-shaded DEM map of the Aareavaara area. Yellow squares mark the position of the oldest Mesolithic sites of the area, Raä 1276
and Raä 1277. Red circles show the position of lake corings. The highest shoreline in this area is ~170 m a.s.l. (marked with blue line), or slightly
below. Thus, most of the map area was covered by Ancylus Lake at the time of deglaciation. The Mesolithic sites are situated between 165 and
170 m a.s.l. Coordinate system according to SWEREF99 TM. © Lantmäteriverket, Gävle, permission 2011/0086.

Möller et al. 107
area and performing a palaeoenvironmental reconstruction of the
consecutive 1000–2000 years. Three lakes were chosen, all situated
below the highest shoreline of the area: (1) Lunkkujävi (~900 m
ESE of site 1276), (2) Aareajärvi (~2.8 km S of site 1276) and (3)
Vähä-Aareajärvi (~3.6 km S of site 1276) (Figure 2). The coring
was conducted with a 7.5 cm diameter Russian peat sampler. The
rods were hammered down to a stop with a lead weight. Sediment
stratigraphy was documented in the field, after which the cores
were transferred to PVC tubes for later sampling and analysis.
Mineral magnetic measurements
The volume specific magnetic susceptibility (κ) of the cores
was measured with a ‘Bartington susceptibility bridge’ con-
nected to an MS2-E1 high-resolution scanning sensor coupled
to automatic measuring equipment. Readings were made every
4 mm along the core with an air measurement between each
measurement.
Age determinations
Radiocarbon dating was carried out on burnt animal bone material
(land mammals) and on plant macrofossils, the latter sieved and
picked from the investigated sediment cores (Table 1). The dated
material was pre-treated in various ways depending on material
type and weight, according to the laboratory protocols of the two
radiocarbon dating laboratories used. Retrieved 14C ages have all
been calibrated into calendar years BP (cal. yr BP) according to the
Oxcal v4.1 package (Bronk Ramsey, 1995, 2001) and given as age
ranges for each dated sample at 1 σ, and also range median age.
Pollen analysis
Sediment cores were sampled over desired depths, in most cases
at 5 cm intervals (2 cm3). Lycopodium tablets were added to allow
quantitative estimates of pollen. Both the pollen accumulation
rate and the percentage values were calculated, the former provid-
ing important information on the quantitative variation for each
Table 1. Radiocarbon dates on macrofossils retrieved from lake cores and burnt bone from archaeological sites Raä 1276 and 1277. LuS
numbers are from the Lund University Radiocarbon Dating Laboratory; Ua numbers are from the Ångström laboratory, Uppsala University.
Locality Depth (cm)
below lake
surface
Dated material Lab. no. 14C yr BP ± 1 σCalibrated age,
1 σ interval
Cal. yr BP
median age
Lake Lunkkujärvi
(164 m a.s.l.)
319–321 19 Pinus needle scales
+ epidermis (bark),
5 bud scales, 1 twig
(2 mm)
LuS 9262 4005 ± 50 4519–4425 4480
(67°26.554′N;
23°32.615′E)
330 1 Pinus needle LuS 9263 4990 ± 55 5876–5650 5725
331–333 24 Pinus needle scales
+ epidermis (bark), 1
Betula fruit
LuS 9509 6515 ± 55 7485–7332 7431
337–339 7 Pinus needle scales
+ epidermis (bark),
1 Betula pub. fruit, 6
bud scales
LuS 9508 7405 ± 70 8327–8174 8239
356–357 Equisetum LuS 9264 8860 ± 65 10,157–9824 9970
365 Equisetum LuS 9265 8900 ± 65 10,170–9919 10,018
370–373 Equisetum LuS 9266 8810 ± 65 10,119–9697 9862
Lake Väha-Aareajärvi
(164 m a.s.l.)
588–590 3 Betula pub. catkin
scales, Salix twigs
(2–3 mm)
LuS 9267 8475 ± 65 9534–9455 9486
(67°24.785′N;
23°31.493′E)
607–608 Equisetum, 1 Betula
pub. catkin scale,
twigs (2–3 mm)
LuS 9268 9100 ± 60 10,371–10,197 10,262
613–615 Equisetum, bark LuS 9269 9320 ± 65 10,650–10,420 10,523
619–621 Equisetum, bark LuS 9270 9395 ± 65 10,708–10,524 10,626
624–626 Equisetum LuS 9271 10,130 ± 65 11,940–11,620 11,767
Lake Aareajärvi (163
m a.s.l.)
512–513 1 Carex, 2 nuts (?) LuS 9513 8775 ± 60 9909–9665 9794
(67°25.345′N;
23°30.875′E)
527–528 1/2 Viola palustris nut LuS 9512 8295 ± 80 9429–9140 9292
537–538 9 Potamogeton fruits,
2 Betula pub. catkin
scales + 1 fruit
LuS 9511 8430 ± 60 9525–9419 9455
Bone datings
Raä 1276
(67°26.651′N;
23°31.162′E)
Burnt bone Ua-38699 10,291 ± 565 12,672–11,216 11,935
Raä 1277
(67°26.685′N;
23°31.163′E)
Burnt bone Ua-38698 9384 ± 488 11,326–9912 10,733
Raä 1276 Burnt bone LuS 9106 8555 ± 60 9554–9485 9530
Raä 1276 Burnt bone Ua-41267 9637 ± 128 11,186–10,786 10,962
Raä 1277 Burnt bone Ua-41266 9192 ± 237 10,741–9947 10,401

108 The Holocene 23(1)
Figure 3. (a) Mesolithic site Raä 1277 in the middle of the picture and, ~50 m behind in the background, site Raä 1276; the archaeological sites
being situated on slightly higher ground on the Koskenkangas sand and gravel plateau. (b) Site Raä 1276, seen from the mire to the south of it.
(c) The plough marks crossing site Raä 1276, where quartz waste and burnt bone fragments initially were discovered. (d) Test pit square 1-2 at
site Raä 1276. Nails in the excavated squares mark finds of quartz and burnt bone fragments. In the upper left and lower right corners are dark
brown stained soil, indicating the outer rim of what is interpreted as a hearth, the central part erased by soil preparation (shallow ploughing).
(e) Quartz flakes as waste products from stone knapping (site Raä 1277). (f) Burnt bone fragments (site Raä 1277). (g) Sediment core from Lake
Vähä-Aareajärvi. All sediment shown was deposited after the basin was isolated from Ancylus Lake owing to isostatic uplift

Möller et al. 109
species, as opposed to the percentage distribution. The samples
were pre-treated according to method A in Berglund and Ralska-
Jasiewiczowa (1986). Near-bottom samples containing minero-
genic sediment were treated with hydrofluoric acid (HF),
sometimes with repeated boiling. Samples were mixed with glyc-
erine and placed on glass and analyzed under the microscope at
400× magnification. As a rule, at least 500 pollen grains were
counted on each level. Calculations and graphs were made using
computer programs Tilia and TiliaView.
Results and interpretation
Archaeology
Lithic material. Most of the finds collected from the surface of
the plough marks are in secondary positions. However, it is still
possible to see clustered distributions in sizes between 3×3 m to
10×3 m. Site Raä 1277 showed two clusters of quartz waste and
one of chlorite phyllite (greenish kind of slate) waste material,
whereas Raä 1276 revealed one cluster of quartz waste. The
quartz occurs as two types, one clear variety and one milky, at
both sites. Within all three quartz clusters there were also clusters
of bone fragments, but none among the chlorite phyllite waste.
The agglomeration of finds indicates separate camp sites, espe-
cially where burnt bones were found.
A total of 61 pieces of quartz were found at Raä 1276, the
assembled weight amounting to ~67 g. Waste material from
quartz knapping included seven flakes made from hand-held plat-
form reduction and two flakes from bipolar reduction, using an
anvil. The rest of the flakes, flake fragments and splinters are all
indeterminable considering reduction technique.
A total of 136 pieces of quartz (Figure 3e) and 40 pieces of
chlorite phyllite were found at Raä 1277, the assembled weight
amounting to ~162 g. Waste quartz material included three cores
or core remnants, positioned within 2 m from each other. One
quartz core was also found in the depression between the two
sites, 25 m from the cores at Raä 1277. Of theses knapping cores,
two are bipolar, one is a hand-held platform core, while one core
is reduced using both reduction strategies. The waste material
from quartz knapping contains seven flakes from platform reduc-
tion and six flakes from bipolar reduction. The rest of the quartz
material is indeterminable with regards to the reduction tech-
nique. The 40 pieces of chlorite phyllite were found in a separate
area at Raä 1277 in which no quartz material was found. There is
thus a clear separation of different stone materials within the site.
No knapping cores of chlorite phyllite were found and, of the five
flakes it was possible to determine, all came from platform reduc-
tion. This greenish material is fine-grained, softer and easier to
shape than quartz, is fragile at the edges of the flakes and easily
broken compared with the edges of quartz flakes. A sustainable
edge of chlorite phyllite must have been ground and polished, but
all the material left on the site was made using percussion only.
Within Raä 1276 and Raä 1277 there are no arrowheads,
scrapers, knives or any sort of finished artefacts, not even second-
arily trimmed flakes. With no finds of finished artefacts it is hard
to make comparisons with similar lithic material from Finland
and Norway. Both bipolar and platform reduction strategies are
common in quartz and do not differ from how quartz material was
processed in the Stone Age in northern Sweden, Norway and
Finland.
Blade and micro-blade production is typical for the oldest sites
in Norway and Finland. Blade production from e.g. Målsnes 1 site
in Norway used any suitable fine-grained material, e.g. Norwe-
gian chert or fine-grained quartzite (Blankholm, 2008). In Raä
1276 and Raä 1277 there are no cores or blades of that kind and
quartz is not suitable for blade production. It should have been
possible to use chlorite phyllite in that way, but no traces were
found. Local raw material seems to have been used, and no
imports were found. Quartz occurs naturally in the whole of
Norrbotten county and there is plenty of naturally occurring
chlorite phyllite at Koskenkangas. Altogether, the sparse lithic
material, its composition and shapes do not indicate the origin of
the first humans coming to Aareavaara.
Bone material. Bone fragments totalling ~10 g were collected
within the Raä 1276 and Raä 1277 sites, all in a more or less burnt
state (Figure 3f). The osteological analyses at microscopic level
of the bone fragments show occurrence of Haversian systems
(osteons), strongly suggesting that they derive from terrestrial
mammals (Leif Jonsson, LJ-Osteology, Gothenburg, personal
communication, 2009). Because of fragmentation (1–5 mm) it
was not possible on an osteological macroscopic level to specify
from which species they derive.
Burnt bone fragments were analysed for identifiable DNA
(Centre for GeoGenetics, Natural History Museum of Denmark,
University of Copenhagen) for possible species identification.
Following DNA extraction of two bone fragments according to
Lorenzen et al. (2011), PCR amplifications were attempted using
generic mammalian primers for mitochondrial 16S (Willerslev
et al., 2003) and primers amplifying part of the d-loop of reindeer
(Lorenzen et al., 2011). Although the expected length of the
amplification products were approximately 100-base-pair includ-
ing primers, no product was obtained with the reindeer primers
and only human and cow DNA (typical contaminants) was
retrieved with the generic mammalian primers following cloning
and sequencing (5–7 clones were Sanger sequenced). Thus, we
conclude that if any endogenous DNA is left in the DNA extract it
must be highly degraded.
From comparison with the Sujula site in Finland (Figure 1;
Rankama and Kankaanpää, 2008) being around the same age and
exposing a total predominance of reindeer (Rangifer tardus L.) as
the hunted animal, this species is obviously the most plausible
candidate for the bone material at the Aareavaara site also, which
unfortunately could not be verified. However, our palaeoecologi-
cal reconstruction (see below) suggests an open woodland birch
tundra at occupation of the investigated sites which further advo-
cates that found bone material stems from reindeer. It is notewor-
thy that there are no bones from fish or from marine mammals in
the material, even though the sites were situated on an island in
Ancylus Lake (see below).
Man-made structures. On each side of the plough mark crossing
the 2 m2 excavated at site Raä1276, were dark, brown-coloured
stains containing burnt bone fragments (Figure 3d). The stains
were shallow, 2–3 cm thick and 5–10 cm wide. The plough mark
was shallow (0.1 m deep), but nevertheless the whole centre of
the structure was gone. Superficial scattered bones and quartz
waste were left. Burnt bones suggest human activity and the
brown stains indicate that it is a man-made structure, but not of
any great depth, e.g. a refuse pit. Burnt bone indicates open fire
and brown-reddish colouring of sand (iron oxide) is produced
from the heat of the fire; we thus suggest that the stained area
marks the former position of a hearth with an oblong shape, mea-
suring ~1.35 m ×0.9 m in size. Edge stones denoting the actual
hearth area were not encountered, but these may not have been
laid out at camp sites occupied only for short periods of time.
Bone clusters were also identified at site Raä 1277, one of which
was partly excavated in a test pit. It did not, however, show stains
or other traces of a man-made structure except for the burnt bone
fragments.
No charcoal was identified within the hearth structure at Raä
1276. The landscape at the time of occupation had a very sparse
occurrence of birch trees (see below). Thus the most probable

110 The Holocene 23(1)
source for making fire would have been shrubs of willow and
dwarf birch. These will burn effectively with good access to
oxygen and will leave only light grey-white ashes blowing with
the wind.
Age and occupancy duration of the sites. In total there are five
radiocarbon age determinations on burnt bone fragments from the
two archaeological sites (Table 1), three of which are on samples
collected during the 2009 survey and two from the 2010 excava-
tions of the test pits at sites Raä 1276 and 1277, respectively.
Burnt bone is often troublesome to date; depending on the degree
of burning there is a variable risk of contamination as the bone
matrix turn into a stable state first at temperatures above 600°C.
At lower temperatures the risk of contamination from penetration
of younger (or older) carbon is evident. The ages of the first per-
formed radiocarbon datings were suspicious, one of them of a
very high age and both of them technically poor because of a very
high age range (≥ ±488 14C years at 1 σ; Ua-38699, Ua-38698,
Table 1). These high standard deviations might be a result of too
small bone samples and not sufficiently thorough pretreatment.
Re-dating of bone fragments from the 2010 excavation turned out
to be technically more accurate; burnt bone from within the iden-
tified hearth at Raä 1276 has an age of 11,186–10,786 (median
age 10,962) cal. yr BP, while the bone cluster identified at Raä
1277 has an age of 10,741–9947 (median age 10,401) cal. yr BP
(samples Ua-41267 and Ua-41266, respectively; Table 1). An
additional dating of burnt bone from the hearth at site 1276 gave
a substantially younger age, 9554–9485 (median age 9530) cal. yr
BP (LuS 9106).
A question thus arises: is the older age set due to contamina-
tion by old carbon, or is the younger age due to contamination by
young carbon? For long distances north to west of Aareavaara
(i.e. upstream of the predominating ice-flow direction) bedrock is
totally devoid of limestone lithologies as a source for limestone
particles in Quaternary sediments, in turn a possible source for
old carbon contamination of burnt bone fragments in surficial
deposits. Instead, bedrock is predominantly granites, metasedi-
mentary rocks (mica schist, quartzite, etc.) and gabbro (Ödman,
1957). An apparent source for old carbon contamination is not
obvious. We thus suspect that the young radiocarbon age (LuS
9106) is due to infiltration of humic acids in porous, poorly burnt
bone. Our preferred interpretation is that the older radiocarbon
ages (except for the oldest one, Ua-38699) are more accurate for
dating occupancy of the Aareavaara archaeological sites, with a
mean age at about 10,700 cal. yr BP.
Lake sediment stratigraphy and chronology
Our sediment core from Lake Aareajärvi (Figure 2) gave too few
macrofossils for construction of a reliable age–depth diagram (the
few retrieved radiocarbon ages also were in reversed order; Table 1),
and was therefore omitted from further studies. Thus the sediment
cores from Lake Lunkkujärvi and Lake Vähä-Aareajärvi have
been used for further analysis of basin isolation and palaeoenvi-
ronmental reconstruction.
Lake Lunkkujärvi. Lake Lunkkujärvi is the smallest of the three
investigated basins, approximately 300 m long, 100 m wide and
with a water depth of only 2.6 m. The lake is situated at about 164
m a.s.l., i.e. ~6 m below the HS and surrounded at short distances
towards the north, west and south by terrain that was above HS at
deglaciation (Figure 2).
A sediment core was recovered just southwest of the basin
centre with the following stratigraphy:
• 273–326 cm: fine detritus gyttja
• 326–348 cm: algal gyttja
• 348–356 cm: clay gyttja
• 356–360 cm: sandy gyttja with abundant macrofossil
remains
• 360–373+ cm: sand (coarse–medium) with abundant mac-
rofossil remains
An age–depth diagram (Figure 4a) was constructed based on
the dated macrofossil remains from seven levels (Table 1). The
three lowermost ages are more or less the same, suggesting very
rapid deposition of the coarse to medium sand and the overlaying
13 cm of sandy gyttja, probably within a few decades. Sedimenta-
tion rate then decreases gradually in the clay gyttja and overlying
units. The age–depth model indicates that the sediments in the
lowermost part of the lake sequence were deposited at around
10,000 cal. yr BP.
The transition from coarse-medium sand to sandy gyttja,
accompanied by stabilizing magnetic susceptibility values (Fig-
ure 5a), could be interpreted as an onset of isolation from Ancylus
Lake. If so, the isolation should have been completed at 3.56 m,
as reflected both in the stratigraphy, with a change from sandy
gyttja to clay gyttja, and a drop in magnetic susceptibility values
(approach zero over a few centimeters). The isolation should then
have been completed in a few decades, shortly after the deposi-
tion of the lowermost layer.
However, a more likely alternative interpretation is preferred:
as the site is situated only ~6 m below the HS and surrounded by
higher areas in most directions it is reasonable to assume that
dead ice was trapped in this shallow basin. In such a case the sedi-
ment succession would not represent an isolation sequence, as
lake sedimentation could only start when trapped dead-ice finally
had melted, which could have taken several centuries. The abun-
dance of macrofossils in the lowermost stratigraphical units sup-
ports this scenario. They indicate that vegetation was already
established around the site when the lowermost units were depos-
ited, which is not compatible with climate conditions at general
deglaciation.
Lake Vähä-Aareajärvi. Lake Vähä-Aareajärvi is a circular basin
with a diameter of about 450 m, located approximately 3.5 km
almost due south of the two archaeological sites Raä 1276 and
1277 (Figure 2). The lake is situated at 163 m a.s.l. and has a
water depth of 3.55 m. Approximately 700 m northeast of the
basin there are a number of areas situated slightly above 170 m
a.s.l., most of which are smaller than 100 m ×100 m. These should
have formed flat islands above the HS at deglaciation. The dis-
tance to larger, continuous areas above the HS are situated 1.5 km
towards the northwest.
A sediment core was recovered just south of the basin centre
with the following stratigraphy:
• 550–608.5 cm: fine detritus gyttja
• 608.5–626 cm: gyttja with sand layers
• 626–633+ cm: medium sand
The four youngest dates (Table 1) are used to construct an
age–depth diagram (Figure 4b). The lowermost dated level
(11,767±167 cal. yr BP at 6.25 m, Lu 9271) differs significantly
from other retrieved ages and must be considered incorrect and
has therefore been excluded in the age–depth model. The lower-
most sand was deposited over a short time at around 10,700 cal.
yr BP, followed by decreasing sedimentation rate at deposition of
overlying gyttja silt with sand layers, as indicated by the two
dated levels in this unit. An even lower sedimentation rate pre-
vailed during deposition of the fine detritus gyttja, according to

Möller et al. 111
Figure 4. (a) Age–depth diagram based on dating of macrofossil remains from six levels in Lake Lunkkujärvi. Error bars represents 1 σ. As
is evident from the diagram the sedimentation started c. 10,000 cal. BP. During the first decades, before the onset of organic production in the
basin and before the catchment was stabilised by vegetation, the sedimentation was rapid before it slowed and stabilised at a significantly lower
and constant level. (b) Age–depth diagram of the sediment succession in Lake Vähä-Aareajärvi. Error bars represent 1σ. The model indicates
that sedimentation started around 10,700 cal. yr BP, which should represent the deglaciation age of the area. It is reasonable to assume that the
medium sand was deposited in a few decades in contact with Ancylus Lake, and that the gyttja silt was deposited after isolation.
Figure 5. Results of the magnetic analyses of the lowermost parts of the sediment successions in cored lake basins. Calibrated radiocarbon
ages according to Table 1. κ represents the volume-specific magnetic susceptibility (dimensionless). (a) Lake Lunkkujärvi. For interpretation see
text. (b) Lake Vähä-Aareajärvi. Based on the stratigraphy, the isolation from Ancylus Lake is placed at 626 cm at the transition from medium
sand to gyttja silt. For interpretation see text.

112 The Holocene 23(1)
the two uppermost dated levels. The sediment succession reflects
a ‘typical’ isolation sequence. The medium sand is interpreted as
deposited in Ancylus Lake. The overlying unit with some organic
production and decreasing sedimentation rate reflects a signifi-
cant change in depositional conditions attributed to deposition in
an almost isolated basin. The presence of sand layers is most
probably the result of storm events when the basin threshold was
transgressed, but could also be the result of unstable soil condi-
tions with subsequent soil erosion as the vegetation was still not
established in the catchment in the newly deglaciated area.
Within three or four decades the soils gradually became more
stabilized because the vegetation cover gradually became
more developed, as a result of improving climate. The deposition
rate slowed even more and the sedimentation changed to pure
organic, fine detritus gyttja, with a very minute minerogenic
component.
The magnetic analysis is more difficult to interpret (Figure 5b).
The assumed isolation event at 626 cm is disguised by very high
magnetic susceptibilities between ~630 and 610 cm. In this inter-
val κ values are an order of magnitude higher than in Lake Lunk-
kujärvi, most probably due to the formation of autogenic greigite.
However, additional magnetic analyses of individual subsamples
are required to further investigate this problem.
The deglaciation of this basin is set at 10,700 cal. yr BP, as
indicated from the onset of sedimentation according to the age–
depth diagram. It is thus not likely that dead-ice was present
here, in contrast to Lake Lunkkujärvi, where sedimentation
seems to have started ~700 years later, interpreted here to be the
result of dead-ice occupancy. This interpretation is also compat-
ible with the fact that Lake Vähä-Aareajärvi is situated in a more
exposed position and at a much greater distance from the HS,
compared with Lake Lunkkujärvi. The rich occurrence of mac-
rofossil remains in the lowermost unit in Lake Lunkkujärvi,
compared with the absence of these in the lowermost strati-
graphic unit in Lake Vähä-Aareajärvi, also speaks in favour of
this interpretation.
Pollen analysis and vegetation development
The results of pollen analysis are presented in Figures 6 and 7,
both as pollen accumulation rates and percentage distribution.
10,700–10,350 cal. yr BP. Estimates of pollen concentration
and accumulation rate show low values in the rapidly deposited
gyttja silt with sand layers in Vähä-Aareajärvi from the time of
deglaciation until about 10,300 cal. yr BP (Figure 6). The
amounts of birch pollen, probably mountain birch (Betula pube-
scens ssp. tortuosa/ssp. Czerepanovii), exceed 500 pollen/cm2
per yr. This suggests that birch was in place shortly after the ice
disappeared from the site. According to Hicks and Hyvärinen
(1999) values between 500 and 1000 birch pollen/cm2 per yr
indicate birch growing at the site, while 1000–1500 birch pol-
len/cm2 per yr indicate presence of a sparse forest. Widespread
forest may not have been present since the area can be consid-
ered as an archipelago in the early phase after deglaciation. The
analysis further indicates that there were plenty of willows
(Salix spp.), dwarf birch (Betula nana), grasses (Poaceae) and
sedges (Cyperaceae). The proportion of grass and herb pollen
is highest in the oldest part of the sediment sequence, which
also indicates open vegetation with only occasional birch trees
(Figure 6). The few alder (Alnus) and pine pollen (Pinus sylves-
tris) found in the oldest parts of the profile are probably long-
distance wind transported and should not be taken as a sign that
these trees had reached the area at this time period. No sedimen-
tation occurred in the Lunkkujärvi basin; this was still blocked
by dead ice.
10,350–9600 cal. yr BP. Pollen concentration and accumulation
rate remained low in the Vähä-Aareajärvi lake basin, rising slowly
after the isolation from Ancylus Lake and up until about 9600
years ago. The low concentration of pollen in the two lowermost
samples from the Lunkkujärvi basin is likely a result of very rapid
sedimentation of sand and sandy gyttja when dead ice melted in
surrounding areas (Figure 7).
The pollen analysis suggests a distinct change in vegetation
above the level corresponding to 9600 years. Both alder and pine
are spreading, which is also seen in the pollen analysis from
Lunkkujärvi. The occurrence of these tree species at about the
same time at both locations can also be taken as an indication that
the chronology of the lake sediments in both Vähä-Aareajärvi and
Lunkkujärvi is consistent. An accumulation rate of pine pollen
between 500 and 1500 grains/cm2 per yr indicates presence of
pine in the vicinity of the sampling site, while >2000 grains/cm2
per yr suggests dense pine forests (Hicks and Hyvärinen, 1999).
The high values of sedges, horsetail (Equisetum) and fern spores
indicate moist conditions. The macrofossil content (sieved for 14C
dating) shows a dominance of horsetail macro remains from the
bottom of the core up to 351 cm depth in Lunkkujärvi, represent-
ing an age of about 9900 cal. yr BP.
9600–4000 cal. yr BP. The level corresponding to approxi-
mately 8600 cal. yr BP in Lake Vähä-Aareajärvi sediments shows
a distinct reduction of fern spores, which may indicate that the
climate then became drier. The forest cover was much denser,
consisting mainly of pine, birch and mountain ash (Sorbus) and,
in damp places, alder. The pollen diagram from Lake Lunkkujärvi
(Figure 7) shows high values of total pollen concentration, pollen
accumulation and tree pollen up until between 6000 and 6500
years ago. They then fall to a significantly lower level. This is
consistent with studies from the Abisko area, showing a maxi-
mum concentration and accumulation of birch and pine pollen up
to that point (Barnekow, 1999; Barnekow and Sandgren, 2001).
After this the climate became colder and wetter compared with
the Holocene temperature maximum, resulting in a gradually
lower treeline, from about 4500 years ago to the present.
Palaeoenvironmental
reconstruction: Conclusions
and discussion
Based on the age–depth sedimentation curves from the two inves-
tigated lake basins, the age of the first lacustrine sedimentation in
each basin can be set. These diverge however, indicating different
deglaciation ages, which is not expected as the distance between
the two lakes is only about 2 km. Lunkkujärvi has a deglaciation
age of ~10,000 cal. yr BP (Figure 4a), while Vähä-Aareajärvi
exhibits a deglaciation age of ~10,700 cal. yr BP (Figure 4b): a
difference of the order of 700 years, which is a considerably long
time in this context. However, the Lunkkujärvi basin is sur-
rounded in all directions but one by terrain that was above the HS
at deglaciation (170 m a.s.l.). Thus our preferred interpretation is
that, because of the closeness to the HS and the shallow water
depth of Ancylus Lake in the area, this depression was occupied
by dead ice at general deglaciation. Thus normal sedimentation
could not start until that ice melted. According to the age–depth
diagram (Figure 4a) this was complete at about 9900 cal. yr BP,
which also means that the Lunkkujärvi basin isostatically was
lifted above the Ancylus Lake level at that time. The registered
clay and algae gyttja is therefore not Ancylus Lake sediment and
isolation from the latter is thus not seen. Otherwise, the age model

Möller et al. 113
before 8000 cal. yr BP is correct, as indicated by the coherence of
the pollen spectra between the Lunkkujärvi and Vähä-Aareajärvi
lakes, e.g. the sharp and synchronous rise in the alder curve at
9600 cal. yr BP; another strong argument for the interpretation of
dead-ice occupancy in the Lunkkujärvi basin.
The stratigraphy in Lake Vähä-Aareajärvi exhibits the
expected sediment succession for a basin with an isolation
sequence from a larger body of water to a local lake basin. The
age of the earliest sedimentation has been set to about 10,700 cal.
yr BP with the age–depth diagram, an age which we suggest cor-
responds to the deglaciation age for the area (Figure 4b). Such a
deglaciation age differs considerably in comparison with that
proposed by Lundqvist (2009: 131); following his ice-margin
recessional lines, an expected deglaciation age of the Aareavaara
area should be in the order of ~10,100 cal. yr BP. The equicesses
of Lundqvist (2009) builds, however, on extrapolations of the
Swedish varved clay chronology from the county of Västerbotten
and northwards, a considerable distance, and is not unchallenged.
Figure 6. (a) Accumulation diagram of pollen and spores from Lake Vähä-Aareajärvi. Only species with values above 20 pollen grains/cm2 per
yr are included. The dashed lines indicate major changes in pollen and spore composition while the solid line represents the isolation from Lake
Ancylus. (b) Percentage diagram of pollen and spores from Lake Vähä-Aareajärvi. The dashed lines indicate major changes in pollen and spore
composition while the solid line represents the isolation from Lake Ancylus.

114 The Holocene 23(1)
Based on 15 dated lake sediment sequences at different altitudes
in the Boden area (~200 km SSW of Aareavaara; Figure 1), Lindén
et al. (2006) proposed a deglaciation age here, and contemporane-
ous formation of the HS, at around 10,500 cal. yr BP. This is ~500
years earlier than indicated by Lundqvist (2009). Aareavaara is on
a recessional line ~100 years older than that through the Boden
area; thus our suggested deglaciation age for Aareavaara (10,700
cal. yr BP) fits well with the results from the Boden area: the 100
year difference in the recessional line age of 10,600 cal. yr BP
is insignificant compared with the 1 σ standard deviations of
performed 14C datings.
The Aareavaara area is located in a transitional zone between
areas in the northeast that were characterized by a subaqueous
deglaciation, i.e. the ice margin was retreating with water at its
front – in this case Ancylus Lake – and, in the southwest, a con-
tinuation of subaerial deglaciation, i.e. dry land in front of the
receding ice margin. Thus here the regional highest shoreline
developed behind an archipelago towards the NE sector. This
Figure 7. (a) Accumulation diagram of pollen and spores from Lake Lunkkujärvi. Only species with values above 20 pollen grains/cm2 per yr
are included. The dashed lines indicate major changes in pollen and spore composition. (b) Percentage diagram of pollen and spores from Lake
Lunkkujärvi. The dashed lines indicate major changes in pollen and spore composition.

Möller et al. 115
consisted of both high- and low-lying islands, the latter quickly
growing in size at the same time as new islands rose above the
water level at progressive relative land uplift. This uplift was
rapid in its early phase (Lindén et al., 2006: figure 6), followed by
a decreasing uplift rate. As Lake Vähä-Aareajärvi is situated at
about 163 m a.s.l., only ~ 7 m or somewhat less below the HS, the
basin that the lake now occupies should have become isolated
from the Ancylus Lake relatively early because of the initially
rapid land uplift. Our lake stratigraphy and age model indicates
that this was finished at about 10,650 cal. yr BP, thus only some
+50 years later than when the area became ice-free and the HS
formed. Because of this short time period when the basin was in
contact with Ancylus Lake, a thicker sequence of Ancylus Lake
sediment did not form; most of the recorded sediment succession
(Figure 3g) in Lake Vähä-Aareajärvi was deposited when this was
already isolated from Ancylus Lake.
The two burnt bone ages from sites Raä 1277 and 1276,
respectively, which we consider most correct (Ua-41266 and
Ua-41267, Table 1) indicate that there was human activity in the
area within the time period from 10,900 to 10,400 cal. yr BP, a
time span which overlaps with the retrieved deglaciation age from
Lake Vähä-Aareajärvi at 10,700 cal. yr BP. It should be noted that
some uncertainty exists, in part because of the age difference
between the two bone ages (representing different periods?),
partly because of the relatively large technical uncertainties in the
datings (standard deviation at 1 σ is 128 and 237 14C years,
respectively). Regardless of this we can conclude that these
hunter/gatherer sites were most probably occupied during a short
time period, humans dwelling here in a pristine landscape over
which the ice just had released its grip. These sites were, both in
time and space, located in close proximity to the retreating ice
sheet margin, but also in a waterfront location on Ancylus Lake as
the Koskenkangas plateau, on which these sites were identified,
formed an island shortly after deglaciation.
Vegetation quickly developed after deglaciation. During the
first few hundreds of years after deglaciation (10,700 to 10,350
cal. yr BP) the area was characterized by open vegetation, includ-
ing occasional birch trees and an abundance of willow and dwarf
birch. The Aareavaara archaeological sites coincide with this
period, and the vegetation reconstruction indicates that there was
a clear wood source for fireplaces in which the burnt bones were
found. The same type of vegetation continued into the subsequent
period (10,350–9600 cal. yr BP), though with a more dense veg-
etation. However, at 9600 cal. yr BP there was a distinct change in
vegetation when pine and alder expanded.
Mesolithic man in a pristine,
newly deglaciated environment of
northern Sweden
The small amounts of quartz waste and burnt bone – even though
our excavated squares (6 m2) and surveyed surroundings at sites
Raä 1276 and 1277 are limited in size – suggest that these were
hunter-gatherer sites where few occupants stayed just for a short
time. A possible number of occupants could have been 3–5 persons,
an estimate corresponding to the size of the pioneer Stone Age
sites described from southwestern Norway (Fuglestvedt, 2005).
As mentioned above, the Aareavaara sites were situated on what
was a small island in Ancylus Lake at time of occupation, proba-
bly more suited for hunting than gathering. However, none of the
bone fragments come from fish or marine living mammals but
most probably from reindeer. In the contemporary or slightly
younger Sujala site in Finland (Rankama and Kankaanpää, 2008;
Figure 1: site 2) the same pattern is seen: no fish bones, despite a
large lake next to the site, but an abundance of reindeer bone (in
combination with some bones from birds). The position of Raä
1276 and 1277 at an edge of the island’s shore bluff with a clear
view over the strait to the – at that time – mainland shore would
have been optimal in terms of hunting animals crossing the water
between the islands. Ethnological and archaeological examples of
the advantage of using water crossings during hunting are known
from e.g. Canada (Gordon, 2003). However, being hunters did not
mean that the humans travelled only by foot; in Aareavaara
10,700–10,600 years ago, boats must have been a necessity for
travelling around in the Ancylus Lake archipelago.
From where did the first people around Aareavaara come?
Having boats, it would have been simple to travel along the Ancy-
lus Lake shoreline from the east, though few contemporary Meso-
lithic sites are known in that direction (the oldest Mesolithic sites
(Figure 1: sites 11, 12) on the Finnish side of Ancylus Lake are
several hundreds of years younger). However, according to
Dulokhanov and Khotinskiy (1984) and Matiskainen (1996) there
are early Mesolithic sites on the White Sea coast and on the Kola
Peninsula (Figure 1), a possible source area with early deglaciation
in the far north. From here it would have been feasible to travel
westwards, reaching the Ancylus Lake coast, and then further
towards the northwest, a total distance from the White Sea coast
of around 400–600 km, and without any glacial ice blocking the
route at that time.
However, with no documented contemporary Finnish Ancylus
Lake shoreline sites to date, it may be more plausible that the
travel route to Aareavaara was through the oldest Mesolithic sites
in northern Norway, all somewhat older than Aareavaara (e.g.
Figure 1: sites 1 and 4; 11,400 and 10,949 cal. yr BP, respectively).
It would have been fully possible to follow the receding ice-sheet
margin southwards towards Aareavaara along the river valleys, a
distance from the closest fjords of about 300–350 km. There are
still uncertainties considering the origin of the first humans
coming to this part of northern Norway, whether they came from
the southwest following the Norwegian ice-free coast or from the
Kola/White Sea area in the east (e.g. Blankholm, 2008; Rankama
and Rankaanpää, 2008). These are questions still to be answered.
Conclusions
In this article we conclude that:
• The oldest reported hunter-gatherer site in northern
Sweden to date has been located to Aareavaara, close to
the Swedish/Finnish border.
• Radiocarbon dating of burnt bone suggests an occupancy
age of ~10,700 cal. yr BP, an age that falls within the age
frame of the oldest ‘Komsa Phase’ settlements along the
north coast of Norway.
• It is not possible to identify the quartz waste material
from knapping at the site into any specific typology, cor-
relative with other early Mesolithic sites in northern
Fennoscandia.
• Our palaeoenvironmental reconstruction suggests that the
site became ice-free ~10,700 cal. yr BP; thus the Aareav-
aara site was, both in time and space, located in close
proximity to the retreating Fennoscandian Ice Sheet mar-
gin, but also in a waterfront location to Ancylus Lake,
forming the highest shoreline at deglaciation of the area
(~170 m a.s.l.).
• In this pristine environment man experienced a landscape
characterized by open vegetation, including occasional
birch trees and an abundance of willow and dwarf birch.
Acknowledgements
Osteological analyses were carried out by Leif Jonsson, LJ-
Osteology, Gothenburg. Dr Jan-Ingolf Kleppe, Tromsø Uni-
versity, provided invaluable literature help in a sometimes

116 The Holocene 23(1)
hard-to-get-through archaeological report jungle. Bengt Göran
Niska, Aareavaara, provided free lodging and gave support dur-
ing archaeological excavations and lake coring. Sincere thanks
to everyone.
Funding
In addition to internal financial support at the County Museum of
Norrbotten and the Department of Geology, Lund University, the
work was funded by a grant from Northland Resources Inc.
References
Barnekow L (1999) Holocene tree-line dynamics and inferred climatic changes
in the Abisko area, northern Sweden, based on macrofossil and pollen
records. The Holocene 9(3): 253–265.
Barnekow L and Sandgren P (2001) Palaeoclimate and tree-line changes
during the Holocene based on pollen and plant macrofossil records from
six lakes at different altitudes in northern Sweden. Review of Palaeobot-
any and Palynology 117: 109–118.
Berglund BE and Ralska-Jasiewiczowa M (1986) Pollen analysis and pollen
diagrams. In: Berglund BE (ed.) Handbook of Holocene Palaeoecology
and Palaeohydrology. Chichester: John Wiley & Sons, pp. 455–484.
Bergman I, Olofsson A, Hörnberg G et al. (2004) Deglaciation and coloni-
zation: Pioneer settlements in northern Fennoscandia. Journal of World
Prehistory 18(2): 155–177.
Bjerck HB (1989) Forskningsstyrt kulturminneforvaltning på Vega, Nordland.
En studie av steinaldermenneskenes boplassmønstre og arkeologiske
letemetoder. Gunneria 61: 1–212.
Björck S (1995) A review of the history of the Baltic Sea, 13.0–8.0 ka BP.
Quaternary International 27: 19–40.
Blankholm HP (2008) Målsnes 1. An Early Post-Glacial Coastal Site in North-
ern Norway. Oxford: Oxbow Books.
Bronk Ramsey C (1995) Radiocarbon calibration and analysis of stratigraphy:
The OxCal Program. Radiocarbon 37: 425–430.
Bronk Ramsey C (2001) Development of the Radiocarbon Program OxCal.
Radiocarbon 43: 355–363.
Clark PU, Dyke AS, Shakun JD et al. (2009) The Last Glacial Maximum. Sci-
ence 325: 710–714.
Dulokhanov PM and Khotinskiy NA (1984) Human cultures and the natu-
ral environment in the USSR during the Mesolithic and Neolithic. In:
Velichko AA (ed.) Late-Quaternary Environments of the Soviet Union.
London: Longman, pp. 319–327.
Fromm E (1965) Beskrivning av jordartskarta över Norrbottens län nedan
lappmarksgränsen. Sveriges Geologiska Undersökning Ca 39, 232 pp.
Fuglestvedt I (2005) Pionerbosetningens fenomenologi: Sörvest-Norge og
Nordeuropa 10200/10000–9500 BP. AmS-NETT 6. Stavanger: Arkeolo-
gisk museum i Stavanger.
Gordon B (2003) Rangifer and man: An ancient relationship. The Ninth North
American Caribou Workshop, Kuujjuaq, Canada, 23–27 April 2001. Spe-
cial Issue. Rangifer 14: 15–28.
Grydeland SE (2002) Nye perspektiver på eldre steinalder i Finnmark – En
studie fra indre Varanger. Viking 2000: 10–50.
Halinen P (2005) Prehistoric Hunters of Northernmost Lapland: Settlement
Patterns and Subsistence Strategies. Iskos 14. Helsinki: The Finnish Anti-
quarian Society.
Hauglid MA (1993) Mellom Fosna og Komsa. En preboreal ‘avslagskultur’ i
Salten, Nordland. Magister thesis, Department of Archaeology, University
of Tromsø.
Hesjedal A, Damm C, Olsen B et al. (1996) Arkeologi på Slettnes. Dokumen-
tasjon av 11000 års bosettning. Tromsø Museums Skrifter XXVI.
Hicks S and Hyvärinen H (1999) Pollen influx values measured in different
sedimentary environments and their palaeoecological implications. Grana
38: 228–242.
Korteniemi M and Suominen E (1998) Nuoliharju W – Suomen vanhin pyyn-
tikuoppa? Studia Historica Septentrionalia 34: 51–67.
Larsen E, Kjær KH, Demidov IN et al. (2006) Late Pleistocene glacial and lake
history of northwestern Russia. Boreas 34: 394–424.
Lindén M, Möller P, Björck S et al. (2006) Holocene shore displacement and
deglaciation chronology in Norrbotten, Sweden. Boreas 35: 1–22.
Lorenzen DE, Nogués-Bravo D, Orlando L et al. (2011) Individualistic species
responses to climate and humans determine Late Quaternary megafaunal
extinction and survival. Nature 479: 359–364.
Lundqvist J (2009) Weichsel-istidens huvudfas. In: Freden C (ed.) Sveriges
Nationalatlas, Vol. Berg och Jord. Stockholm: Almqvist & Wiksell Inter-
national, pp. 124–135.
Mangerud J, Gyllencreutz R, Lohne Ø et al. (2011) Glacial history of Norway.
In: Ehlers J Gibbard PL and Hughes PD (eds) Quaternary Glaciations
– Extent and Chronology – A Closer Look. Developments in Quaternary
Sciences 15, Elsevier, pp. 279–298.
Matiskainen H (1996) Discrepansies in deglatiation chronology and the
appearance of man in Finland. In: Larson L (ed.) The Earliest Settle-
ment of Scandinavia and its Relationship with Neighbouring Areas. Acta
Archaeologica Lundensia, No. 24, Stockholm: Almquist & Wiksell Inter-
national, pp. 251–262.
Ödman OH (1957) Beskrivning till berggrundskarta över urberget i Norrbot-
tens län. Sveriges Geologiska Undersökning Ca 41, 151 pp.
Östlund O (2004) Rapport. Arkeologisk förundersökning. Stenålders-
boplats samt skärvstensförekomst. Raä 22 samt Raä 98, Junosuando
sockan, Norrbottens län, Västerbotten. Luleå: Norrbottens museum,
dnr 442-2004.
Rankama T and Kankaanpää J (2008) Eastern arrivals in postglacial Lapland:
The Sujala site 10000 cal BP. Antiquity 82: 884–899.
Svendsen JI, Alexanderson H, Astakhov VI et al. (2004) The Late Weichse-
lian Quaternary ice sheet history of northern Eurasia. Quaternary Science
Reviews 23: 1229–1271.
Tikkanen M and Oksanen J (2002) Late Weichselian and Holocene shore
displacement history of the Baltic Sea in Finland. Fennia 180(1–2):
9–20.
Willerslev E, Hansen AJ, Brand T et al. (2003) Diverse plant and animal
DNA from Holocene and Pleistocene sedimentary records. Science 300:
791–795.
Woodman PC (1993) The Komsa Culture: A re-examination of its position in
the Stone Age of Finnmarka. Acta Archaeologica 63: 57–76.