warmlander zori byock scott 2010 hrisbru metals,

[ Pobierz całość w formacie PDF ]
Journal of Archaeological Science 37 (2010) 2284
e
2290
Contents lists available at
Journal of Archaeological Science
journal homepage:
Metallurgical
ndings from a Viking Age chieftain
’
s farm in Iceland
Sebastian K.T.S. Wärmländer
a
,
b
,
*
, Davide Zori
c
, Jesse Byock
d
, David A. Scott
e
a
Department of Anthropology, University of California in Santa Barbara, Santa Barbara CA 93106, USA
b
Division of Biophysics, Arrhenius Laboratories, Stockholm University, 106 91 Stockholm, Sweden
c
The Cotsen Institute of Archaeology, University of California in Los Angeles, Los Angeles CA 90095, USA
d
The Cotsen Institute of Archaeology & Scandinavian Section, University of California in Los Angeles, Los Angeles CA 90095, USA
e
The Cotsen Institute of Archaeology, UCLA/Getty Conservation Programme, University of California in Los Angeles, Los Angeles CA 90095, USA
article info
abstract
Article history:
Received 2 January 2010
Received in revised form
5 April 2010
Accepted 9 April 2010
The metalworking, metal import, and use of metal in medieval Iceland is still little understood. When the
Scandinavian settlers colonized Iceland in the 9th c. AD, the island was found to contain no useful metal
deposits save for bog iron, and the deforestation that followed the settlement resulted in a scarcity of
wood. Only in the last decades have archaeological excavations begun to unravel how the
rst Icelanders
dealt with this lack of resources. This paper presents the metallurgical
ndings from a Viking Age
chieftain
s present-day capital
Reykjavik. The excavated metal objects had all been crafted with good workmanship employing tech-
nology similar to that used in mainland Scandinavia. However, most excavated metal
’
s farmstead at Hrísbrú in the Mosfell valley, located just outside Iceland
’
Keywords:
Archaeometallurgy
Corrosion
Conservation Science
Viking studies
X-ray diffraction
X-ray
nds show
evidence of re-use, which together with the second-grade metal in some of the objects indicates
a shortage of raw material that prompted the Icelandic colonizers to improvise and make do with
whatever material was at hand.
Even though this chieftain
uorescence
s farm was materially poorer than contemporaneous high-status farms in
mainland Scandinavia, it was wealthy by Icelandic standards. The analytical results show that some
excavated objects were imported trade goods deriving from both neighboring and far-away localities,
proving that the farm was part of the extensive trade network of the Viking world. Most likely, this farm
represents the upper limit to what a Viking Age farm in Iceland could afford in terms of material objects
and trade goods.
’
2010 Elsevier Ltd. All rights reserved.
1. Introduction
nds excavated from
an Icelandic chieftain
’
s farm, dating to the earliest days of the
Icelandic settlement. By comparing the material culture of this farm
with contemporaneous Icelandic and Scandinavian sites, an
example is provided of how a relatively wealthy Icelandic farm
managed to compromise between the Scandinavian customary use
of metals and Iceland
This paper presents the analysis of metal
While the history of metalworking in mainland Europe goes back
to the earliest times, metalworking in Iceland only began when the
island was colonized by the Scandinavian settlers in the 9th century.
The settlers found Iceland to be a land of volcanic rock, containing no
useful metal deposits save for bog iron. All other metals, particularly
copper, silver, and gold, had to be imported. The deforestationprocess
that followed the settlement eliminatedmost of Iceland
’
s lack of metals and other material resources.
s native forest
of dwarf birch (Betula L.) and willow (Salix sp.) (
Samset, 1991
, 27), and
it has been argued that ironproductionwas a driving force behind this
tree-cutting (
Smith, 1995
). However, the scope and nature of the
metalworking, the metal import, and the metallurgical technology in
the early days of Iceland
’
2. The Hrísbrú site
At the Hrísbrú site, located in the Mosfell valley just a few
kilometers outside Iceland
’
s
capital Reykjavik,
the Mosfell
Archaeological Project has excavated a 10
11th century farmstead
(
Fig. 1
). The Mosfell valley runs from the bay Leirvogur in the west
to the highlands of the Mosfell heath in the east. The excavated site
is located along the slope of the northern side of the valley, at
a strategic position from which it is possible to view both the
central valley area and the coastline to the west (Byock et al., 2005).
So far, a traditional Viking Age (ca. AD 790
e
’
s history still remain to be fully understood.
Corresponding author at. Division of Biophysics, Arrhenius Laboratories,
Stockholm University, 106 91 Stockholm, Sweden. Tel.:
*
þ
46 8 162447.
e
1100) longhouse, a farm
E-mail address:
(S.K.T.S. Wärmländer).
0305-4403/$
e
see front matter
2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2010.04.001
 S.K.T.S. Wärmländer et al. / Journal of Archaeological Science 37 (2010) 2284
e
2290
2285
Fig. 1.
Map showing the Mosfell Valley and the Hrísbrú excavation site in Southwest Iceland.
church with an associated cemetery, and a pagan cremation site
have been unearthed.
Viking Age Scandinaviawas a strati
plowing, and the stratigraphy of the soil is mostly intact up to the
topsoil, even though the meadow has been mechanically leveled in
modern times. For example, above the longhouse, a volcanic ash
layer from the Katla eruption in AD 1500 was found in situ some 20
cm below the current surface level.
In all excavation areas, i.e. the longhouse, the church, the
surrounding graves, and the cremation site, metal artifacts were
encountered. Although the artifacts were heavily corroded, some
contained enough remaining metal to allow extraction of useful
metallurgical information.
ed societywith a chie
yelite
(
Byock, 2001
,66
e
69;
Roesdahl, 1999
), and
Vésteinsson (2004
,
74
75) has grouped Icelandic houses into low, middle, and high
status by reference to three parameters, i.e. house size, artifacts/
prestige goods, and historical records indicating status. At Hrísbrú,
the longhouse measures an impressive 28m fromend to end, which
makes it one of the largest Viking Age longhouses so far excavated in
Iceland. The excavation yielded more imported glass beads than any
other archaeological farmstead in Iceland (see below), and there are
several medieval Icelandic sagas mentioning chieftains living at the
old Mosfell farm, which was located at the site of modern-day
Hrísbrú (
Byock et al., 2005; Grímsson, 1886
). Thus, there is ample
evidence for Hrísbrú being one of the more important high-status
households in Iceland during the 10
e
11th centuries.
Although the Hrísbrú farmstead has been continually inhabited
from the Icelandic settlement up to the modern period, the habi-
tation sites have changed over time, and the excavated Viking Age
structures are located no more than 50 m from the present-day
farmhouse. The longhouse dates from the original settlement of
Iceland in the late 9th to early 10th century, and the small
accompanying church was built around AD 1000. Both the long-
house and the church are very well preserved as no subsequent
structures or occupation has disturbed the site since it was aban-
doned in the 11th or early 12th century, save for a small agricultural
building located on top of the knoll where the church is located. The
e
3. Materials and methods
uorescence (XRF) spectroscopy was carried out on all
metal samples in order to characterize their elemental composi-
tion. The XRF spectra were recorded with a tungsten lter at 45 kV/
11 mA, and with no
X-ray
lter at 35 kV/10 mA, using a Jordan-Valley
Excalibur benchtop XRF (model EX-2600U).
X-ray diffraction (XRD) spectra were recorded of corrosion
products and bead colorants using a Rigaku R-Axis Spider unit
employing an image plate in Weissenberg geometry to collect
Debye-Scherrer-rings. Minute amounts of material were scraped off
the samples and placed on the end of a rotating glass spindle, and
XRD spectra were recorded at 50 kV/40 mA for between 600 and
1500 s. After baseline-correction, the spectra were searched and
matched against reference spectra from the International Centre for
Diffraction Data (ICDD) using the JADE v8.2 software fromMaterials
Data Inc.
Fourier transform infrared (FTIR) spectra were recorded with
a Perkin
oor of the longhouse was buried approximately 1 m below the
surface of what is now a meadow, where the soil covering the
longhouse has been deposited through domestic trash dumping
inside the cavity of the house, followed by aeolian soil deposition
from the nearby eroding mountains. This aeolian soil is slightly
acidic (pH 6
Elmer Spectrum One instrument equipped with a solid
state Attenuated Total Re
e
ectance (ATR) sample stage. Spectra of
corrosion products were recorded in the 4000
550 cm
1
region
with a resolution of 4.0 cm
1
and matched against the UCLA/Getty
Conservation Program
e
7) and generally sandy, allowing rainwater to seep
through and access any buried objects. The average rainfall in the
valley is about 100 mm/month, and the average temperature in the
valley ranges from
e
’
s IR database, as well as the spectral data-
base of the Infrared and Raman Users Group (IRUG).
Small samples were cut from some of the metal
2
C in the winter to
14
C in the summer,
þ
nds with
creating yearly freeze
thaw cycles (personal communication from
Guðrún Gísladóttir at the Icelandic Meteorological Of
e
ajeweler
s saw, mounted in Buehler transparent Epoxide resin, and
oriented to expose the cross-sections. The samples were then ground
and polished in steps ending with the Buehler Metadi 1 micron dia-
mond suspension. The polished cross-sections were etched in 2%
ferric chloride for copper and 2% nital for iron in order tomake visible
the grain structure in the samples (
Scott, 1991
). The etched surfaces
were examined under re
’
ce, 2008).
The preservation conditions at the site are therefore unfavorable,
and most excavated objects are severely deteriorated due to the
exposure to a combination of oxygen and moisture. On the upside,
it appears that the meadow inwhich the structures are situated has
been used exclusively for grazing, never for growing crops.
Consequently, the buried material has not been disturbed by
ectedaswellaspolarizedlightinametal-
lographic stereomicroscope at 50
e
400
magni
cation.
2286
S.K.T.S. Wärmländer et al. / Journal of Archaeological Science 37 (2010) 2284
e
2290
4. Results and discussion
Together with two pieces of iron slag (2007-21-273; -294) found on
the longhouse
oor, the hammer scales suggest that small-scale
iron working, such as repairs, took place at the site. Iron making
using local ore has been demonstrated at different sites in Iceland
(
Espelund, 2007
), such as in the nearby Reykjavik area where
a number of 9th-10th century smithies and iron bloomeries have
been found (Vala Garðarsdóttir, personal communication, 2010),
but so far no evidence for large-scale iron production has been
found at Hrísbrú.
The cemetery included a number of iron artifacts, consisting
mostly of heavily corroded nails and clench bolts from burial
features. The iron nails and bolts had remnants of wooden planks
attached, and as these planks are otherwise completely decom-
posed, it appears that the iron in the nails and rivets has slowed the
disintegration of the adjoining wood. The clench bolts consist of
round-headed nails which have been passed through boards and
then hammered over an iron square or diamond-shaped rove
placed around the nail (
Fig. 3
A). In Scandinavia, this type of clench
bolt was typically used in the construction of ships and smaller
boats built in the klinker style, and is also known as a ship rivet (
Bill,
1994; Lundström, 1972; McGrail, 2004
)(
Fig. 3
B). In the burials, the
iron nails were found along the edges of the burials, often in
associationwith rectangular organic soil stains from decayed wood,
indicating that they were part of cof
4.1.
Iron objects
Iron artifacts were unearthed from the longhouse, the church,
the cemetery and the cremation site.
In the longhouse four knife blades (2006-27-48; 2007-21-106;
-107; -110) were found on the
ttings
(2006-27-49; -50) belonging to a wooden construction. All four
knives were single-edged with straight backs, and measured
around 10 cm in length. No handles remained, suggesting that
these were made from wood or bone which has now decomposed.
XRF measurements showed no presence of other metals in the
knife blades, indicating they had been crafted without decorations
of e.g. silver, tin, or bronze/brass. As both the knife blades and the
iron
oor together with two iron
ttings were completely mineralized without any pseudo-
morphic retention of microstructural detail in the iron corrosion
products, further metallographic analysis was not possible.
In the church building, a knife blade (2003-41-46) was found in
a stratigraphic layer above the church
oor, post-dating the aban-
donment of the church. The blade retained some solid metal under
a thick crust of corrosion, allowing a sample to be taken for metal-
lurgical analysis. The cross-section revealed that the blade was
fashioned from two pieces of wrought iron, welded around a core of
high-carbon steel, and aligned along the length of the blade (
Fig. 2
).
The ne pearlite between the acicular ferrite plates allows the
carbon content to be estimated at about 0.2
ns. The clench bolts, which are
unsuitable for making square boxes, were found on top of the
skeletons, indicating that they were part of wooden constructions
other than cof
0.3% carbon, and it has
been known since the early days of iron-working that suchmixing of
softer iron with harder steel can improve the overall material
properties of iron objects (
Carpenter and Robertson, 1930a, 1930b
).
Unfortunately, not enough of the cutting edge has survived to make
it possible to tell whether the steel was quenched in this area or not.
The particular fashioning of this knife closely matches the three-
layered
e
nd clench bolts or
ship rivets in burial features, and it appears that the clench bolts
and the decayed wooden planks that they bound together are
recycled material from boats (
Zori, 2007
). Re-use of material from
Viking Age boats in Christian burials has previously been reported
from Sebbersund in Denmark, although there the iron
ns or burial biers. It is unusual to
ttings had
been removed for re-smelting and the boards were found below
the bodies, suggesting to
Birkedahl and Johansen (1995)
that the
re-use was purely functional. At Hrísbrú, the presence of the bolts
and their placement on top of the body precludes a purely utili-
tarian role of these objects, and suggests that these boat fragments
were ritually deposited in the graves as symbolic references to the
pagan tradition of boat burials and the ship as a vehicle to the
afterlife (
Zori, 2007
).
In addition to the clench bolts, two other iron grave goods were
found in the cemetery. In burial Feature 3, a nine cm long hooped
iron dress pin was unearthed (2002-57-55) (
Fig. 4
). In mainland
Scandinavia, dress pins were used by both men and women from at
least the migration period (ca. AD 370
method (
Tylecote and Gilmour, 1986; Tylecote,
1987
, 263), which has been employed in mainland Europe since at
least the
“
sandwich
”
rst centuries AD (
McDonnell, 1989; Ottaway, 1992
). This
method does not appear to have reached Scandinavia until the
9th
10th centuries (
Arrhenius,1970,1989a; Peets,1995; Lyngstrøm,
1995
). Upon its arrival, it gained wide-spread popularity, and soon
became the dominant method for steeling knives in Scandinavia
(
Arrhenius, 1998, 1989b; Carlsson, 2003; Lyngstrøm, 1995
).
Underneath the church
e
oor, in a stratigraphic layer related to
an earlier structure, small magnetic metal
akes (2003-41-74) were
encountered. XRD spectra of the
ed them as magne-
tite, Fe
3
O
4
, indicating that they are magnetic hammer scales,
formed at high heat during smithing activities (
Tylecote, 1987
, 320).
akes identi
e
570) onwards, and in the
Vendel period (ca. AD 570
e
790) the early protuberant style was
Fig. 2.
Etched cross-section of knife-blade 2003-41-46, found in the church building at the Hrísbrú farm. A) The rightmost region is wrought iron, while the leftmost region is
carbon steel. The region in between is the weaker welded interface, displaying cracking and corrosion. B) This close-up of the leftmost region in (A) shows the typical Wid-
manstätten structure for low-carbon steel.
S.K.T.S. Wärmländer et al. / Journal of Archaeological Science 37 (2010) 2284
e
2290
2287
Fig. 3.
A.) Examples of the clench bolts found in burial feature CK-2007-5 in the graveyard next to the Hrísbrú farm church. Note the decomposed wood that has been preserved due
to close proximity to the corroded iron. B.) Clench bolts employed in two methods for joining overlapping planks (redrawn by Jennie Dillon from Ottaway, 1992, 617).
replaced by polyhedral pins (
Waller, 1996
). In the Viking Age, the
use of dress pins declined, as women began to fasten their clothing
with characteristic oval brooches (
Jansson, 1985
). Men, on the other
hand, continued to use dress pins which were usually made from
bronze or brass, but sometimes they were fashioned from iron, and
occasionally from expensive silver (
Waller, 1996
). Pins with hoops
were typically used in pairs, which allowed a chain to be strung
between them that could be used for carrying decorative or utili-
tarian items. A pin made from iron had multiple purposes, as it was
sturdy enough to also be used as a tool for punching holes or as an
awl. For
incomplete dress pin pair was deposited in a cemetery including
partial boat material. Presumably the excavated parts were meant
to symbolically represent complete objects.
4.2. Bronze fragments
A handful of copper alloy objects and fragments were encoun-
tered at the Hrísbrú site, and four fragments were selected for
detailed technical examinations, i.e. 2007-21-61 found in the
collapsed southern wall of the longhouse, 2003-41-33 from the
church
nd 2002-57-55 only one of the hooped pins, and not the
full pair, was deposited in the grave. The corroded state of the pin
precludes stylistic and metallographic analysis. To the best of the
authors
oor, 2002-57-68 from the cremation site, and 2006-27-16
from the soil above the longhouse. The 2006-27-16 fragment rested
within the tephra layer from the Katla eruption of AD 1500,
approximately dating the
knowledge, it is the only iron dress pin so far encountered
in Iceland.
The second iron gravegood is a fragment of an iron pot (2003-
41-65). It corresponds to about 1/5 of the original vessel, and was
found deposited in burial Feature 49. The diameter of the original
vessel was estimated to 19 cm, and the fragment consists of two
iron sheets welded together with a loop riveted to the pot (
Fig. 4
).
The latter allowed a chain to be fastened so that the pot could be
suspended over a
’
nd to that year. Due to their fragmentary
and corroded nature, it is not possible to accurately determine the
original alloy composition of the
nds or fromwhat kind of objects
they derive. However, X-ray uorescence analysis revealed that the
four
nds consist of tin-bronze alloys. This constitutes a deviation
from the standard composition of Scandinavian Viking Age copper
alloys, which typically contain zinc, often lead, and less frequently
tin (
Arrhenius, 1989c; Craddock, 1990; Oldeberg, 1966; Söderberg,
2010
). As there are no tin deposits in Scandinavia, geographic
proximity suggests that the tin might originate from the British
Isles where tin has been mined since at least 1000 BC (
Barton, 1957;
Varyl et al., 2004
), even though other regions of origin are possible
also. For 2006-27-16, XRF analysis identi
re. The design and construction is similar to
other welded and riveted iron pots from Viking Age Scandinavia,
such as the ones found as grave goods in pagan graves (
Fridell, 1930
,
229;
Odencrantz, 1937
). The typical pagan grave however encom-
passes whole objects, while at Hrísbrú a fragment of a pot and an
ed small amounts
Fig. 4.
A) Hooped Viking Age iron dress pin (2002-57-55) of West-Norse/Celtic style, found in Burial Feature 3 at Hrísbrú in the Mosfell Valley. B) Fragment of an iron pot (2003-41-
65), found in the same cemetery. Bar is 1 cm. C) Reconstruction of the iron pot. The curvature of the fragment suggests that the pot was rather small, measuring about 19 cm in
diameter. Bar is 1 cm.
2288
S.K.T.S. Wärmländer et al. / Journal of Archaeological Science 37 (2010) 2284
e
2290
(around 2% each) of lead and silver in the material, in addition to
the copper and tin. While lead and tin are common additions to
copper, silver is not, suggesting that the fragment was manufac-
tured from a recycled bronze object with some silver decoration.
Metallographic analysis revealed annealing twins, small grain
sizes, and occasional strain lines in the microstructure of all
samples, indicating that cycles of cold-working and annealing were
employed to skillfully hammer the objects into their desired
shapes. In samples from objects 2002-57-68 and 2006-27-16
relatively high amounts of
common in the Viking world, and while bead 2006-27-53 was
excavated in pristine shape, formation of silver sul
de (Ag
2
S) on
bead 2007-21-179 has caused all the silver surface to turn golden,
suggesting differential preservation conditions at the site.
For the three yellow beads 2006-27-9, 2006-27-54, and 2007-
21-204, XRF analysis indicated the presence of tin, in addition to
the normal elements found in glass such as lead, silicon, calcium,
and potassium. XRD spectra of minute scrapings of bead 2006-27-
9identi
ed the compound lead-tin oxide, PbSnO
3
, also known as
the pigment lead-tin yellow (
Fig. 5
). This pigment comes in two
chemical forms, where type I, Pb
2
SnO
4
, is more common than type
II, PbSnO
3
. Both forms are chemically stable and lightfast, which
helps to explain the pristine appearance of the beads. Lead-tin
yellowwas used in European easel paintings from the 14th century
onwards (
Kühn, 1968
), but its main use during the Middle Ages
was as colorant in yellow glass (
Estaugh, 2004
). Venice was the
center in Europe for such glass production (
Estaugh, 2004
), which
makes Italy a plausible origin for the raw material in the beads.
However, as glass colored with lead-tin yellow type II has recently
been encountered at local glass production sites outside Italy, such
as the Merovingian site of Schleitheim in Switzerland (
Heck et al.,
2003
), the geographic origin of the glass cannot be absolutely
ascertained.
de inclusions were
encountered, indicating that the copper in these two objects was
not of the highest quality, even though the copper must have been
processed with some sophistication in order to yield sul
copper
sul
de rather
than oxide inclusions. In 2006-27-16, cracks were present in
addition to numerous strain lines, providing evidence of heavy
stress-related deformation. Possibly the object had a functional
rather than a decorative use
e
perhaps as a hinge
e
and the
deformation may have been related to this use.
4.3. Beads
A number of monochrome and polychrome glass beads found in
the longhouse were subjected to technical analysis and found to
contain different metallic colorants. In mainland Scandinavia, bead-
making from recycled glass or imported glass rods has been
documented at sites such as Ribe, Hedeby, Kaupang, and Birka
(
Callmer, 1977; Sode, 2004
). Since no glass-working sites have been
found in Iceland,
4.4. Corrosion and deterioration
Due to the environmental conditions at the site all metal
nds
it is safe to assume that these beads were
were heavily corroded. X-ray diffraction identi
ed the copper
corrosion as standard copper oxide, Cu
2
O, and copper carbonate,
(CuCO
3
$
imported in their
nished form.
Four of the beads (2007-21-142, -143, -144, and -145) have an
intricate design pattern, consisting of a black body adorned with
white bands together with eyes in red and blue (
Fig. 5
). XRD and
XRF analysis indicate that the white colorant is tin oxide (SnO
2
), the
red colorant is haematite (Fe
2
O
3
), the black colorant is magnetite
(Fe
3
O
4
), while the blue colorant is an unidenti
Cu(OH)
2
). For iron standard oxides and hydroxides, i.e. FeO,
Fe
2
O
3
, and FeOOH, were found together with the more uncommon
iron carbonate, FeCO
3
, which previously has been encountered in
wet environments such as Danish peat bogs (
Matthiesen et al.,
2003
). For the iron (2002-29-87) and bronze (2002-29-68) frag-
ments deriving from objects deposited in association with the
cremation burial (
Fig. 6
), no unusual corrosion products were
found, and Raman spectroscopy of the corrosion crust of the iron
fragments identi
ed copper-based
compound. Beads with the same characteristic design pattern have
been found at other places in Iceland (
Eldjárn, 1956
)aswellasin
Birka, Sweden, (
Arbman, 1940
), and in Eastern Europe (
Callmer,
1977
, 97). The other nd sites all date to a narrow time window
of AD 960
ed incorporated particles of graphite (i.e. char-
coal) originating from the cremation event, which explains the
charred appearance of the pieces.
Due to the uniform thickness of the bronze fragments from the
cremation site it was possible to calculate the corrosion rate at the
site. The cross-section of fragment 2002-57-68 contains a core of
solid metal in a corrosion matrix retaining the shape and dimen-
sions of the original bronze sheet (
Fig. 6
). With the average thick-
ness of the studied fragment being 200
990 (
Callmer, 1977
, 85), which seems to be a plausible
date also for the current
e
nds from the Hrísbrú longhouse. It
appears that beads with eyes were never produced in Scandinavia,
and
Callmer (1977)
suggests that the style of the four excavated
beads originated in Turkmenistan. Most likely, these beads were
imported along the Viking Age trade routes that descend from
Scandinavia down the rivers of Russia towards Asia.
Two of the other beads, 2006-27-53 and 2007-21-179, were
manufactured with a decorative metal foil coating, identi
10
m
m, and the average
thickness of the remaining metal being 20
3
m
m, the corrosion
ed with
XRF as silver. This technique for bead ornamentation was relatively
layers on both sides measure roughly 90
m. With radiocarbon
dating of charred twigs providing an approximate date of AD 990
m
Fig. 5.
A) One of the four
(F-2007-143) found in the Hrísbrú longhouse, stylistically provenanced to central Asia. B) The yellow bead F-2006-9. C) XRD spectrum
identifying the yellow colorant in F-2006-9 as PbSnO
3
, i.e. lead-tin yellow type II. (For interpretation of the references to colour in this
“
eye-beads
”
gure legend, the reader is referred to the web
version of this article.)
[ Pobierz całość w formacie PDF ]