You are on page 1of 13

M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

Available online at

Archaeometallurgical investigation of the iron anchor from the

Tantura F shipwreck
A. Aronsona , D. Ashkenazia,, O. Barkaib , Y. Kahanovb

Faculty of Engineering, Tel Aviv University, Ramat Aviv 69978, Israel

Leon Recanati Institute for Maritime Studies, University of Haifa, Haifa 31905, Israel



Article history:

The Tantura F shipwreck was a coaster or a fishing vessel about 15.7 m long, discovered in the

Received 1 November 2012

Dor/Tantura lagoon, Israel in 1995. It was dated to between the mid-7th and the end of the 8th

Received in revised form

centuries CE. Among the finds excavated were two T-shaped type iron anchors. Of the two

6 January 2013

anchors, one (anchor A) was thoroughly studied by archaeometallurgical methods in order to

Accepted 14 January 2013

identify forge-welding lines, to determine the welding quality and to understand the
manufacturing technology. The examinations included X-ray radiography, XRF analysis,


optical microscopy, SEM/EDS observation and analysis, OES analysis and microhardness tests.


The investigation included characterization of the composition, microstructure, thermal


treatments, forge-welding junctions and slag analysis. The results revealed a heterogeneous


microstructure, rich in glassy, fayalite and wstite slag. Iron based phases included ferrite,

Tantura F

pearlite, cementite and Widmansttten plates, all typical to wrought iron. The forge-welds of

T-shaped anchor

Anchor A were located. Each arm was made of one piece, weighing about 2.53 kg and the

Wrought iron

shank was made of a few 1.52 kg pieces. The second anchor (anchor B) was only briefly
examined visually and with a few radiographs, which support the results from anchor A.
The research results revealed significant information about T-shaped anchors and their
manufacturing process, including hot-working processes without any additional heat
treatments, and folding techniques. The microstructure was similar to other ancient simple
tools such as saws, sickles, axes and mortise chisels, and though the technology to make
complicated structures and objects, such as swords, existed at that time, the anchors did not
require this sophistication; thus simpler techniques were used, presumably because they were
more cost-effective.
2013 Elsevier Inc. All rights reserved.



The Tantura F shipwreck is the remains of a coaster or a fishing

vessel about 15.7 m long. It was discovered in 1995, about 70 m
offshore in Dor (Tantura) lagoon, which is located on the
Mediterranean coast of Israel, south of the ancient Tel Dor,
about 25 km south of Haifa. It was found in about 1 m of water
covered by about a 0.85 m layer of sand, the thickness of which
changes due to sea conditions. Tantura F was excavated during

five seasons in 20042008 [13]. Based on both 14C and ceramic

typological analysis, the shipwreck was dated to between the
mid-seventh and the end of the eighth centuries CEthe local
early Islamic period [2]. Among the finds exposed in the
shipwreck site were two T-shaped iron anchorsanchor A
and anchor B (Fig. 1). Anchor A was found beneath the hull,
touching it (Fig. 1a), while anchor B was found concreted
to the external part of the planking remains below the hull
(Fig. 1b). The anchors were found covered by a thick grey layer of

Corresponding author at: School of Mechanical Engineering, Tel Aviv University, Ramat Aviv, Tel Aviv 69978, Israel. Tel.: +972 3 6405579;
fax: +972 3 6407617.
E-mail address: (D. Ashkenazi).
1044-5803/$ see front matter 2013 Elsevier Inc. All rights reserved.

M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0


Fig. 1 The Tantura F shipwreck anchors covered with thick encrustation and concretion: (a) anchor A and (b) anchor B.

encrustation and concretion composed of sea sand, shells and

small stones. Both were broken at the shank, with part of the
shank and the anchor cable ring missing.
The present study is part of a systematic series of
investigations concerning the Tantura F shipwreck (e.g., [13]).
According to Eliyahu et al. [3], the connection of the anchors to
the wreck of Tantura F is based on their location at the site and
their typology, considering the date of the shipwreck. Anchor
A's typology was similar to anchors B and to other similar
anchors such as the anchors found in the Yassada I shipwreck
[4]. The anchors were under the ship, which rules out the
possibility that they were dropped after the wrecking of Tantura
F. If the anchors had been dropped by a vessel earlier than the
Tantura F, they would probably have sunk deeper into the sand,
as a result of the activity of the sea, including waves, currents
and sand movement inside the lagoon.
The anchors were removed from the seabed to the laboratory:
Anchor A in the third season (2006), and anchor B in the fourth

season (2007). After documentation and X-ray radiography,

the concretion layers were removed carefully, while watching
the radiographic images in order to cause minimum damage to
the metal remains of the anchors. The cores of the artifacts
were revealed: despite the heavy concretion and the oxidation
process during 1300 years underwater, the metal survived in
a good state of preservationa hard core of iron was evident
(Fig. 2 for anchor A). The arms of both anchors had a rectangular
cross-section and were not precisely perpendicular to the shank.
Their tips were flattened and the cross-sections of both shanks
were circular, which is unusual for a wrought iron object, but
characteristic of T-shaped iron anchors [3].
A wide variety of anchors in shape and material were used
in antiquity, late antiquity, Byzantine (local), Early Islamic
(local) and Middle Ages. They were developed according to
the technological knowledge, capabilities, and experience, as
well as requirements, considering the size of the vessel, and
the nature of the ground, weather and water conditions [5].

Fig. 2 Anchor A: (a) covered with a thick concretion, (b) illustration of the iron remains, relative to the concreted artifact
(estimated superposition, not a real cut) and (c) after the concretions layers were removed.


M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

A typologicalchronological analysis of iron anchors, based on

anchors discovered up to the 1980s, was suggested by Kapitn
[6]. The earliest type of iron anchor had straight arms in
V-shaped position and was dated to Roman Republican times.
The earliest T-shaped anchors to be discovered were excavated in the Dramont D shipwreck, France, dated to the first
century CE and the Dramont F shipwreck, dated to the second
half of the fourth century CE [69]. The last version of iron
anchors (in Kapitn's typology) was the Y-shaped anchors
which are dated to the 10th and 11th centuries [10,11].
However, T-shaped anchors were found in the amaltiburnu
shipwreck assemblage, dated to the thirteenth century [12].
According to Kapitn, typologicalchronological development of T-shaped anchors over the years can be distinguished
[13]. The first T-shaped anchors had a rectangular cross-section
shank and stock, and they appear to be a transition between the
Roman and the Byzantine types. The second type is characterized by long straight flukes and relatively short arms, bent
slightly upwards, and a round hole in the shank for a detachable
stock (similar to the early anchors discovered in the Dramont F
shipwreck). The latest type, which was discovered in the
Yassada I shipwreck, dated to CE 625, is characterized by
upward-curved flukes bent outwards, longer arms and a round
hole for the stock in a rounded cross-section shank [4].
In summary, the use of T-shaped iron anchors was common
throughout the Mediterranean, Marmara and Black Seas over a
period of approximately 1200 years, starting from the first
century and lasting to the thirteenth century. At the beginning
of this period, they were present in parallel to curved armed
anchors, while from the tenth century they appeared together
with Y-shaped anchors. However, it is difficult to establish a
detailed typological development.
Analysis of one T-shaped iron anchor from the Yassada I
shipwreck revealed that it was forged-welded, almost clear
of slag, had a carbon content of 0.07 wt.% and small amounts
of other elements: Co (1.0 wt.%), Ni (0.50 wt.%), Zr (0.50 wt.%),
K (0.06 wt.%), Si (0.14 wt.%), Al (0.05 wt.%), Cr (0.10 wt.%), S
(0.01 wt.%), Na (0.01 wt.%), Mg (0.01 wt.%), Ca (0.01 wt.%), Ti
(0.01 wt.%), V (0.01 wt.%) and Mg (0.01 wt.%) [3,14].
Analysis of the manufacturing process of anchors by welding
about 16 pieces of iron, the largest weighing 56 kg, is described
in the report of the Y-shaped anchors of the eleventh century
Sere Liman shipwreck [10]. Analysis of the iron is given as well
In a previous study of the anchors of Tantura F, the forgewelding lines could not be detected [3]. It was therefore
suggested that further work should be performed, including
detailed radiography and further metallographic examination
of sections from other parts of the anchors, as well as
extended element slag inclusion study.
The present study is thus an extended investigation of the
Tantura F anchors. It aims to determine the manufacturing
processes of the anchors, focusing on the forge-welding
processes of anchor A and identifying the locations of the
forge-welding lines and their quality. The importance of this
research is due to its comprehensive metallurgical and
chemical analysis which was done over the entire anchor A.
A significant part of this work was the development of a
research methodology for investigating large archeological
artifacts that were suspected to have been assembled from a

Fig. 3 The straight scarf weld and the cleft weld [20].

number of wrought iron blooms that were forged welded

together. The research development methodology is based on
knowledge that has been borrowed from forensics science and
failure analysis methods.
Since these anchors are dated between the mid-seventh
and the end of the eighth centuries CE, it is logical that other
T-shaped anchors from the same period were manufactured
using similar methods. Thus, the present analysis can serve
as a tool in future studies of similar artifacts.


Background of the Metallurgical Analysis


The History of iron Manufacturing and Processing

Since the melting point of pure iron is around 1540 C, which

can be reached only with nineteenth century technology,
ancient iron was produced at lower temperatures in a reducing
atmosphere using the smelting process. The main method for
smelting iron ores was the direct or bloomery method, in
which the ore was inserted into a furnace with carbon-rich fuel,
at temperatures up to 1200 C. In this process, the iron never
reached the melting point, but as iron oxides reduced, the ore
coalesced as a solid mass of metal called the bloom and the
slag was separated as a melt. Then the slag-rich sponge was
subsequently hot worked (forged) to squeeze the slag out and to
consolidate the iron, producing wrought iron [16]. According to
furnace limitations, the ancient bloom was usually small and
in order to form a ship's anchor, several small blooms, a few
kilograms each, were forge-welded together [17].
The basic furnace was a cylindrical clay shaft, with
detentions of between 1 and 2 m in height, an internal diameter
of between 0.3l m and walls normally over 0.2 m thick, to
reduce heat loss and air inlets, called Tuyeres or blowing holes,
which were located about 0.30.5 m from the furnace base.
The commonest form of fuel was charcoal. In most furnaces, an
arch through the wall of the furnace enabled the slag to be
removed, either cold or as tapped slag.

M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0


Fig. 4 Images of T-shaped anchors from the Tantura F shipwreck: (a) anchor A, (b) anchor B.

At the beginning of the smelting process the furnace was

first charged with fuel, then preheated. When it was hot, a
mixture of ore and charcoal was inserted into the furnace and
air was pumped into it. The iron ore was initially reduced
inside the furnace at temperatures of around 800 C whereas
slag was formed near the base of the furnace at temperatures
over 1000 C. The major reactions occurring during smelting
are: (1) the reduction of iron oxide to metallic iron, and (2) the
formation of a liquid slag. During the first reaction, carbon
monoxide (CO) was formed by a reaction between the carbon
present in the fuel and the oxygen in the air. During
the second reaction, liquid slag, such as wstite (FeO) and
gangue oxides (silica, alumina, etc.), was formed. The slag was
separated from the metal by allowing the liquid to slap to the
base of the furnace and then by removing it. If the furnace
temperature was not high enough, the separation was
not completed successfully and the bloom was composed of
metallic iron mixed with slag. The blooms were often
heterogeneous, varying in composition from pure ferritic
iron, phosphoric iron (up to 1 wt.% P) to carbon steels
(containing up to 0.8 wt.% C). In order to reduce the amount
of included slag, the bloom was broken up by hammering,
separating the ductile iron from the rest of the material.
The hammering removed the majority of the semi-fluid slag
and oxides and then, the iron particles were welded together
to form a bloom iron, producing iron with some slag [18].
The bloom iron could be then worked up into different iron
objects by its smelter or sold to another blacksmith.
Iron smiths had different alloys available to them, including
pure and soft ferritic iron, harder and more brittle phosphoric
iron, and steels with varying carbon contents, enabling the
fabrication of hard and tough edges. The main techniques used

by the smith were cold working, hot working, forged-welding

and heat treatments. In cold working (also known as work
hardening and strain hardening), the metal is deformed at a low
temperature, strengthening the metal by plastic deformation
which displaces dislocations and slightly hardens the iron.
In hot working the metal is deformed above its recrystallization
temperature (over 600 C in the case of iron), enabling the smith
to easily shape the metal. Forged-welding joins two metal parts
together by heating them to high temperatures and hammering
them together. In heat treatments, the metal is heated to
improve specific properties. For example, steel can be made very
hard (and brittle) by heating it to 900 C and then quenching it in
cold water. However heating the metal increases the possibility
that the metal will oxidize and become useless. This can be
prevented by controlling the fire and by fluxing the metal surface
with sand [19].


Forge-welding Techniques

Forge-welding (also called fire welding) is a process whereby

two pieces of wrought iron are joined together by a combination
of thermal and mechanical methods (such as hammering),
without the addition of a solder. When wrought iron is heated
up to 1350 C, it can be easily forged-welded by hammering.
The process must be completed before the iron joint cools
below 1050 C [18]. The iron silicate in the slag, left over from
the smelting or added during sand fluxing [19], facilitates
the welding by inhibiting corrosion, which is faster at high
temperatures [18].
There were two basic techniques for welding two parts
together: the straight scarf weld and the cleft weld [20], as
illustrated in Fig. 3. Using these techniques, the welded contact


M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

Fig. 5 Anchor A: (a) sectional plan (cylindrical hole at the centre

of the anchor was made by Eliyahu et al. [3] during the
preliminary study), (b) parts of anchor A after cutting.

area is bigger and the pressing force from hammering is more

effective, resulting in better forge-welding [21]. Therefore, in
the case of connecting two blooms that were shaped as rods,
such as the shank and the arms of the anchors (that were
constructed from several pieces), it is expected to find similar
forge-welding forms of jointsstraight scarf or cleft weld.

Previous Study of the Iron Anchors from the Tantura
F Shipwreck
Eliyahu et al. [3] previously studied the two anchors (Fig. 4),
but only analyzed two different zones of each anchor. These

zones were analyzed by typological and archaeometallurgical

methods, including radiography, metallographic cross-sections,
SEM with EDS analysis, Optical Emisson Spectroscopy (OES)
analysis and microhardness tests. They found a heterogeneous microstructure in both anchors, consisting of ferrite,
Widmansttten ferrite-pearlite plates and pearlite, which
are typical phases and morphologies in wrought iron
made by the bloomery process. Combining their OES and
SEM/EDS results, they suggested that the anchors had a similar
composition. Soda-blast cleaning followed by chemical
etching revealed lines suggesting the presence of forgeweldments [3].
Eliyahu et al. [3] concluded that since both anchors had the
same minor elemental inclusions (except for phosphorous
that appears in the inclusions from Anchor A only) and since
analysis of minor and trace elements present in individual
inclusions should reflect the elements present in the original
ore, the two anchors were probably manufactured by the
same process and from the same ore. They assumed that a
poor quality weld could be detected by radiography, whereas a
good forge-weld might not be easily detected by radiography
nor observed by an Optical Microscope (OM) and Scanning
Electron Microscope (SEM). Since the radiography images
showed no sign of welding lines in both anchors, and since
the representative metallographic images showed no forgewelding lines, and no unique amount of oxide or glassy
inclusions were observed in all sections, they concluded that
the T joints of the anchors had probably been welded very
professionally by a very highly skilled blacksmith who was
able to avoid bulk defects in the forge-welding process. In
addition, the metal survived in a very good state of preservation: the shank was found with a core of about 5 cm
diameter of unoxidized iron. This might have been a result of
environmental conditionsthe anchors of Tantura F had
been buried under a layer of sand about 1 m thick. Accordingly, they assumed that the only way to observe good
forge-welding is by slightly etching the surface of the anchors
and then observing the result. Eliyahu et al. [3] suggest that
taking into account the weights of the artifacts and the macro
forge-weld lines, with regard to the manufacture of both
anchors, it may be suggested that anchor A and B were made
from at least four pieces.


Fig. 6 Forge-weld lines at the throat (shank-arm junction).

Experimental Methods and Tests

Since anchor A had less corrosion than anchor B, the

present research focused on anchor A, identifying the
locations of its forge-welding lines and evaluating its
welding quality. Anchor A was analyzed by its typology
and by archaeometallurgical methods, including radiography, Stereo-microscopy (SM), metallurgical Optical Microscopy
(OM), SEM (Jeol JSM-7000F field emission SEM, with a voltage
of 25 kV), Energy Dispersive Spectroscopy (SEM/EDS), X-ray
fluorescence spectrometry (Thermo Scientific Niton XL3T 900
handheld XRF analyzer, with a geometrically optimized large
area drift detector technology), Optical Emission Spectroscopy
(Baird HR-400 instrument), Vickers microhardness test. Anchor
B was analyzed only by its typology and by radiography
(Table 1).


M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0

Table 1 Analyses performed on each anchor.











Anchor A

Anchor B





Fig. 7 Radiograph image of the: (a) area in the anchor which was radiographed, (b) higher magnification of the radiograph revealing
the forge-weld line (arrows).

X-ray radiography was performed in different locations and

projections of both anchors in order to reveal the locations of the
forge-welding lines and the welding quality, using a GE-ERESCO
MF3 300 Kv X-ray tube, with 200250 kV working voltage,
3.54.5 mA working current and exposure time of 2.54.5 min.
The dimensions and mass of both anchors were measured
in order to estimate from how many pieces the anchors
were fabricated and how many junctions were forged-welded.
Anchor A was cut into pieces using a water-cooled steel disc.
Eight metallographic cross-sections, A1A1 up to A8A8,
were taken from the different zones of anchor A (Fig. 5) and

sections were prepared according to the ASTM E3-01. The

sample surfaces were ground with silicon carbide 240 to 600
grit papers, then polished with alumina paste from 5 to 0.05 m,
and then polished with 0.05 m colloidal silica suspension
pastes. To remove contaminants, the samples were cleaned
with ethanol and dried. Next, the samples were etched
with Nital (97 mL ethyl alcohol and 3 mL nitric acid). The
metallographic samples were examined under an OM (ZEISS,
AXIO-Scope A.1). Vickers microhardness tests were performed
using a Future-Tech Model FM-700e microhardness tester, with
a load of 100-g force.

Fig. 8 Anchor A cross-sections after cutting: (a) section A1A1, (b) section A2A2.


M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

Fig. 9 OM photograph of Widmanstatten plates (A1A1





Anchor A

The shank of anchor A had a round cross-section; its arms had

rectangular cross-sections and were almost perpendicular
to the shank, as typical for T-shaped anchors [3,5,22]. There
were visual indications over the surface of the anchor, suggestive
of forge-welding connections, such as shown in Fig. 6.
More than ten radiography tests were performed on anchor
A revealing a suspected forge-welding area that was not seen on
the surface (Fig. 7). The suspected area was characterized in the
radiographs by a tilted inner planar surface, about 4.5 cm along
the shank, starting about 5 cm from the arms. It should be noted
that the darker areas in the radiographs indicate absence of
matter. Examination of the outer surface of the shank revealed
no signs of an underlying weldment. It should be noted that
the radiographs did not cover the entire anchor surfaces.
Complementary sections were also made.


Section A1A1

OM observation of the metallographic samples of section A1

A1 (Figs. 5a, 8a) after etching revealed a heterogeneous
microstructure of ferrite, pearlite and cementite network as
well as Widmansttten plates (Fig. 9). Microhardness measurements on section A1A1 revealed two areas with typical
average hardness values of 154 25 HV and 219 41 HV.
The two areas were separated by a large cavity (Fig. 8a).

Table 2 XRF analysis results from two measurement





Compositions weight percentage (wt%)












Fig. 10 OM photograph showing heterogeneous microstructure of cementite (center), pearlite (left) and
Widmansttten plates (right).
The carbon content was estimated from the measured
microhardness to be 0.20.6 wt.%.
XRF measurements were performed at two different points
on section A1A1, as summarized in Table 2. The composition
was found to be mostly Fe, 98.899.5 wt.%, with some other
elements, including up to 0.1 wt.% Cr, up to 0.7 wt.% Al,
0.1 wt.% P and between 0.3 and 0.4 wt.% Si (Table 2).


Section A2A2

Macro examination of external surface of A2A2 revealed a

fine line that might indicate the forge-weldment that was
seen in the radiographs. Examination of section A2A2 after
cutting it (Figs. 5a, 8b) revealed an inner cavity whose size,
location and orientation agreed with the radiographs, indicating forge-welding. OM and SEM observation of the metallographic samples of section A2A2 after etching revealed a
heterogeneous microstructure of ferrite, pearlite, cementite
network and Widmansttten plates (Fig. 10), as well as wstite
(FeO) slag inclusions. The measured microhardness was
between 85 HV, which is typical for pure iron, up to 248 HV,
which is typical for pearlite, with average value of 123 70 HV.
Two areas were analyzed with SEM-EDS at a magnification
of 50, as summarized in Table 3a. It was found that the
composition was 95.997.3 wt.% Fe, 2.33.4 wt.% C and small
percentages of Al, Si, P, S, Mn and As.


Section A3A3

Observation of the metallographic samples of section A3A3

(Fig. 5a) with OM and SEM revealed a heterogeneous microstructure of ferrite, pearlite, cementite, and Widmansttten plates, as
well as slag of wstite (FeO) and glass. The microhardness was
Table 3a SEM-EDS elemental analysis of the metal.

A2A2 (area 1)
A2A2 (area 2)

Compositions weight percentage (wt%)
















M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0

Fig. 11 SEM photos of: (a) a glassy slag inclusion, (b) three-phase (wstite-fayalite-glass) slag inclusion surrounded by -ferrite
grains, (c) and (d) two-phase slag (wstite-glass) surrounded by -ferrite grains.

between 77 HV (typical for pure iron), up to 123 HV, which is

typical for ferrite with some pearlite, with an average value
of 10414 HV. The slag microhardness was 487582 HV. No
evidence of forge-welding lines was identified in this section,
specifically not between the arm and the fluke.


Section A4A4

OM and SEM observation of metallographic samples of

section A4A4 (Fig. 5a) after etching revealed a heterogeneous
microstructure of ferrite, pearlite, cementite and Widmansttten
plates, as well as a high level of slag inclusions such as wstite,
fayalite and glass. The microhardness was between 78 HV and
169 HV (average hardness of 9925 HV), which is typical for ferrite
with a small amount of pearlite. The slag hardness was 273493
The composition of the metallic region, determined by
EDS, was 98.5 wt.% Fe, 0.8 wt.% C and other elements (Si, Mn

and O), as summarized in Table 3a. The EDS composition of the

of glassy slag (Fig. 11a) was 26.7 wt.% Fe, 19.7 wt.% Si, 4.1 wt.% Al,
5.4 wt.% Ca and of 30.3 wt.% O, and smaller amounts of Mg, K,
Mn, Mo, Ti (Table 3b). Three-phase slag, wstite-fayalite-glass,
and two-phase slag (wstite-glass) in ferrite iron was also
observed by SEM-EDS (Fig. 11b, c, respectively). The two-phase
slag was composed of 43.279.9 wt.% Fe, 0.113.9 wt.% Si and of
17.429.8 wt.% O, and small amounts of Al, Mn and Ca (Table 3b).
XRF indicated an iron composition of 99.3100.0 wt % Fe,
with up to 0.1 wt.% P and up to 0.6 wt.% Si (Table 2). OES
indicated that the composition included Fe, C, Si, Ni, Mo, Ti
and P (Table 4).


Section A5A5

Macro-examination of section A5A5 (Fig. 5a) revealed no signs

of forge-welding. Therefore, this section was not examined

Table 3b SEM-EDS elemental analysis of the slag.


Fig. 11a
Fig. 11c point
Fig. 11c point
Fig. 11c point
Fig. 11c point

Compositions weight percentage (wt.%)























































M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

Table 4 OES analysis (average of 3-points).




Compositions weight percentage (wt%)













Section A6A6

Macro examination of the external surface of A6A6 (Fig. 5a)

revealed two forge-welding lines (Fig. 12, white arrows). Observation of the metallographic samples of section A6A6 with OM
after etching revealed a heterogeneous microstructure of ferrite,
pearlite, cementite network and Widmansttten plates, as well
as transparent glassy slag, two-phase slag (wstite-glass) in
ferrite iron and two-phase slag (fayalite laths in glass) in a ferrite
matrix (Fig. 13). Microhardness measurements made for the A6
A6 section revealed an average value of 15339 HV, which is
typical to a combination of ferrite and pearlite. The slag hardness
values are between 544 and 673 HV.


Section A7A7

The goal of examining section A7A7 (Fig. 5a) was to determine

whether the observed external layer was a coating layer at the
external part of the anchor or whether it an oxidation layer.
SEM/EDX analysis of the external layer revealed only a high
percentage of O and Fe without any other elements, confirming
that there is no outer layer or coating.


and the metallurgical findings shown in Fig 15. The location

of this area, close to the throat also supports its being a forgewelding (Fig. 15a). The direction of the slag, shown in Fig. 15b,
is parallel to the borderline, supporting the interpretation of a
forge-welding line here.
OM and SEM-EDS examination and analysis around and along
the borderline in the metallographic sample of section A8A8
(Fig. 15b) after etching did not reveal strong evidence of slag, flux
usage or oxidation lines. The change in the microstructure at the
forge-welding line was continuous and smooth, indicating high
quality forge-welding in this area (Fig. 15).


Anchor B

As has been mentioned, anchor B was studied to a limited

extent in comparison to anchor A. Visual examination of
anchor B revealed that its shank was round and its arms had a
rectangular cross-section and were almost perpendicular to
the shank, similar to anchor A (Fig. 4). There were visual
indications on the surface of anchor B, including the throat,
suggesting forge-weldments, such as shown in Fig. 16.
Only a few radiographs were made on parts of the surface of
anchor B. In the radiography photographs that were performed
on anchor B, no evidence of forge-welding lines was seen,
probably as a result of significantly greater corrosion damage
than in anchor A. The significant corrosion damage made it
difficult to distinguish between the deep surface corrosion and
inner cavities that might be a result of imperfect forge-welding.

Section A8A8

Macro examination of the external surface around the throat

(the junction between the shank and the arms) revealed
indications of forge-welding connection lines, as shown in
Fig 6. Metallographic examination of section A8A8 (Fig. 5a)
after etching revealed clear and sharp borders between two
different microstructures (brighter and darker), one with ferrite
grains and the other with pearlite grains, separated by a region
with Widmansttten structure.
The clear and sharp border suggests that two parts were
forge-welded together, each with a different microstructure.
The tilted view in Fig 14 shows the connection between the
indications of forge-welding lines on the external surface

Fig. 12 Tilted view of section A6A6 showing two dimensional

joining plain (arrows).



Multidisciplinary typological and archaeometallurgical methods

were used to investigate the manufacturing processes of the two
anchors, focusing on anchor A. The purpose was to characterize
the composition, microstructure, forge-welding junctions and
slag inclusions. Dozens of metallurgical cross sections were
studied in different locations and orientations.
The archaeometallurgical analysis of Anchor A indicates that
it was made of wrought iron. It had a heterogeneous multi-phase
microstructure of ferrite, pearlite, cementite and Widmansttten
plates, rich in slag (glassy slag, fayalite, wstite), as typical of
wrought iron. The parallel elongated wstite slag indicates that
the anchor was manufactured by hot working. Although silica
slag particles were found, there is no hard evidence that silica
was specifically used as a flux in the forge-welding areas.
The metallographic investigation revealed that anchor A was
manufactured without any additional heat treatments and
folding techniques, probably because of cost-effectiveness
considerations. In the majority of the cross-sections, variable
carbon zones were observed, from low carbon zones (<0.3 wt.%)
to medium (0.30.6 wt.%) and high carbon zones (>0.6 wt.%). The
hardness of anchor A is similar to that of anchor B [3, p. 241] and
matches that of other working tools such as ancient wrought
iron saws, sickles, axes and mortise chisels [23].
From both radiographic and metallographic observation it
was seen that anchor A was of high quality, and manufactured
by expert blacksmiths. Only the combination of radiographs,
visual observation, OM and SEM/EDS observation and analysis

M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0


Fig. 13 OM photos of fayalite (Fe2SiO4) slag matrix: (a) surrounded by ferrite and pearlite, (b) high intensity of light revealing
the slag texture, (c) higher magnification of the ferrite and pearlite, and (d) higher magnification and high intensity of light, revealing
the slag texture.
of the metallographic sections made it possible to locate the
forge-welds of anchor A and examine them. It was found that
the straight scarf weld method was used to join the anchor
blooms. Each arm of anchor A (the arm and the fluke) was made
of one 23 kg unit, while the shank was made from a few
1.52 kg blooms (at least five blooms, Fig. 17). During the first
manufacturing stage, the shank and the crown (lower part of
the shank) round sections were manufactured and combined.
The arms were manufactured separately, and shaped to their
rectangular shaped cross-section with the flukes at the end of
the arms. At the next stage, each arm was connected to the
shank to form the final anchor.

The results from anchor B, although briefly studied, show

that each arm was made from two pieces, each piece weighing
about 3 kg. The remaining piece of the shank weighs about
7 kg and most probably was composed of at least two pieces,
each one weighing 34 kg.
This assumption is based on the total weight and measurements of the arms and the shank remains and on the forge
welding location assumption according to the visual indications
that were observed over the surface of the anchor.
Beside iron and carbon, XRF, SEM-EDS and OES showed the
presence of Si, Al, Ti, Cr, Mn, As, P, Mo and Ni in anchor A iron,
while SEM-EDS analysis of the slag revealed Fe, C, Al, Si, Mg, K,

Fig. 14 Tilted view revealing evidence of forge-welding joint in section A8A8.


M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

Fig. 15 (a) forge-weld joints in section A8A8; (b) SEM image showing border between two different microstructures ferrite
and pearlite, separated by a Widmansttten microstructure zone.

Mn, O, Ca, Mo and Ti. Comparing our results with the chemical
analysis of the anchor from the Yassada I shipwreck [14]
indicated that most elements were detected in both anchors.
This study also corresponds to the metallurgical analysis of the
iron anchors from Yassada I that reveal the use of wrought iron
with similar microstructure and composition [14] to those of
anchor A. In the Sere Liman shipwreck, the anchors were
composed of several blooms, but the welding lines were not
detected during the metallurgical examination [10,15].
Good identification of the forge-welding locations must
combine visual inspection, radiography and metallurgical
sections, together with some prediction and understanding
where forge welding connections can be expected. This work
showed that each one of the inspection methods alone is
not enough to identify all the locations. The combination of
non-destructive and destructive testing enabled us to reach
good understanding and enriched our knowledge in all that is

concerned with manufacturing processes of anchors made in

the local early Islamic period.



Among the finds excavated from the Tantura F shipwreck,

two T-shaped type iron anchors were discovered, dated to
between the mid-seventh and the end of the eighth centuries
Anchor A was made of wrought iron, with heterogeneous
microstructure, manufactured by hot working and then its
pieces were joined together by straight scarf forge-weld.
Anchor B generally supports the results from anchor A,
but being larger, its arms were made from two blooms each.
Although the technology of making complicated structures,
e.g., used in swords, using heat treatments and folding

M A TE RI A L S C HA RACT ER I ZA TI O N 78 ( 20 1 3 ) 1 0 81 2 0


Fig. 16 Anchor B: (a) photograph suggesting forge-weld joints (arrows), (b) 3-D illustration schematic showing the different parts
of the anchor (top view).

techniques, existed at that time, anchors did not require

this sophistication and so simpler techniques were used,
presumably because they were more cost-effective. The anchor

microstructure was similar to other simple ancient tools such

as saws, sickles, axes and mortise chisels.
Since the typology of the anchor studied here is similar to
other anchors from the same period, our understanding of the
manufacturing and forge-welding techniques may contribute to
the understanding of the manufacturing processes of those

This research was supported by the Israel Science Foundation,
the Hecht Foundation, Lord Jacobs of London, a Sir Maurice
Hatter Fellowship for Maritime Studies and the University of
Haifa, to whom the authors are thankful.
We gratefully thank Izhak Hershko and Dan Brightman from
the radiography department of the SOREQ Nuclear Research
Centre, for their outstanding and constructive help.
Many thanks go in particular to Shirly Druker, Demitry
Fishman, Igal Galper, Ofer Levi, Ulia Naim, Gil Shemesh,
Efrat Shwartzberg and Din Wilson.


Fig. 17 Forge-welding lines revealed in Anchor A () and

suggested reconstructed forge-welding lines ( ).

[1] Barkai O, Kahanov Y. The Tantura F shipwreck, Israel. Int J

Naut Archaeol 2007;36(1):2131.
[2] Barkai O, Kahanov Y, Avissar M. The Tantura F shipwreck
The ceramic material. Levant 2010;42(1):88101.
[3] Eliyahu E, Barkai O, Goren Y, Eliaz N, Kahanov Y, Ashkenazi D.
The iron anchors from the Tantura F shipwreck: typological
and metallurgical analyses. J Archaeol Sci 2011;38(2):


M A TE RI A L S CH A RACT ER IZ A TI O N 78 (2 0 1 3 ) 1 0 81 2 0

[4] Van Doorninck Jr FH. The anchors. In: Bass GF, van Doorninck
Jr FH, Yassi Ada, editors. A seventh-century Byzantine
shipwreck, Volume I. College Station: Texas A&M University
Press; 1982. p. 12141.
[5] Kemp P, editor. The Oxford companion to ships and the sea.
London: Oxford University Press; 1976. p. 21.
[6] Kapitn G. Ancient anchorstechnology and classification. Int J
Naut Archaeol 1984;13(1):3344.
[7] Joncheray J-P. tude de l'pave Dramont D: les objets
mtalliques. Cah Archeol Subaquatique 1975;4:518.
[8] Joncheray J-P. Une epave du Bas Empire: Dramont F. Cah
Archeol Subaquatique 1975;4:91140.
[9] Haldane D. Anchors in antiquity. Biblical Archaeology
[10] Van Doorninck Jr FH. The anchors. In: Bass GF, Matthews SD,
Steffy JR, van Doorninck Jr FH, editors. Sere Liman an EleventhCentury Shipwreck, Vol. I, The Ship and Its Anchorage, Crew, and
Passengers. College Station: Texas A&M University Press; 2004.
p. 189233.
[11] Pulak C. The wrecks of Yenikapi the Gift of Storm. Arkeoatlas
[12] Gnsenin N. A 13th-century wine carrier: amalti Burnu, Turkey.
In: Bass G, editor. Archaeology Beneath the Seven Seas. London:
Thames and Hudson; 2005. p. 11823.
[13] Kapitn G. Exploration at Cape Graziano, Filicudi, Aeolian
Islands 1977. Int J Naut Archaeol 1978;7(4):274.
[14] Delwiche DE. The anchors. In: Bass GF, van Doorninck Jr FH,
Yassi Ada, editors. A seventh-century Byzantine shipwreck,
Vol. I. College Station: Texas A&M University Press; 1982.
p. 3224.

[15] Stech T, Maddin R. The anchors. In: Bass GF, Matthews SD,
Steffy JR, van Doorninck Jr FH, editors. Sere Limani an
eleventh-century shipwreck, Vol. I, the ship and its
anchorage, crew, and passengers. College Station: Texas
A&M University Press; 2004. p. 1925.
[16] Pleiner R. Iron in archaeology: the European Bloomery smelters.
Praha: Archaeologicky U'stav Aver; 2000.
[17] Buchwald VF, Wivel H. Slag analysis as a method for the
characterization and provenancing of ancient iron objects.
Mater Charact 1998;40:7396.
[18] Jobling HJW. The history and development of English anchors
C.A. 1550 to 1850. College Station TX: Texas A&M University;
[19] McDonnell G. Iron working processes. The historical metallurgy
society: archaeology datasheet no.3. J Hist Metall Hist Metall Soc
[20] Anon. The Blacksmith's craft: an introduction to smithing for
apprentices and craftsmen. Great Britain: Countryside Agency
Publisher; 1955.
[21] Sims L. The backyard blacksmithtraditional techniques for
the modern smith. Gloucester, Mass: Quarry Books; 2006.
[22] Galili E, Dahari U, Sharvit J. Underwater surveys and rescue
excavations along the Israeli coast. Int J Naut Archaeol
[23] Scott DA. Metallography and microstructure of ancient and
historic metals. Marina del Rey; Santa Monica: Getty
Conservation Institute; 1991.