JAAS

PAPER

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article OnlineView Journal

Non-destructive

aInstitute of Archaeology, Research Centre

Sciences, Uri Street 49, 1014 Budapest, HunbInstitute of Historical Sciences, University

HungarycWigner Research Centre for Physics, Hungar

Konkoly-Thege Miklos Road 29-33, HungarydCentre for Energy Research, Hungarian Acad

Thege-Miklos Road 29-33, HungaryeWosinsky Mor Museum, 7100 Szekszard, Sz

Cite this: DOI: 10.1039/c4ja00377b

Received 30th October 2014Accepted 15th December 2014

DOI: 10.1039/c4ja00377b

www.rsc.org/jaas

This journal is © The Royal Society of

analyses of bronze artefacts fromBronze Age Hungary using neutron-based methods

V. Kiss,*a K. P. Fischl,b E. Horvath,c Gy. Kali,c Zs. Kasztovszky,d Z. Kis,d B. Marotid

and G. Szaboe

In this paper we present the application of various neutron-based methods carried out at the Budapest

Neutron Centre. Non-destructive and non-invasive neutron radiography (NR), prompt gamma activation

analysis (PGAA) and time-of-flight neutron diffraction (TOF-ND) analysis were applied to reveal more

information on raw material and production techniques of bronze artefacts that can be dated to the

Central European Bronze Age (2000–1200 BC).

1. Introduction

Various neutron-based methods (neutron radiography,prompt gamma activation analysis and time-of-ight neutrondiffraction; Szentmiklosi 2013) were applied at the BudapestNeutron Centre, Hungary, thanks to our EU FP7 NMI3 pilotproject (Studies on local metal production of the CarpathianBasin from the late Copper Age until the Middle Bronze Age3500–1500 BC) in order to explore the compositions and themanufacturing techniques of copper and bronze artefacts.Most of the ndings are unique in their form and function;that is why, considering aspects of heritage protection, wedecided to use non-destructive methods to provide moreinformation on raw material and production techniques. Inthis paper we present the investigation of three bronzeobjects from the Middle and Late Bronze Age in Hungary(2000–1200 BC) (Fig. 1).

2. Analytical methods

Various non-destructive and non-invasive neutron-basedinstruments of the Budapest Neutron Centre (with thecooperation of the Wigner Research Centre for Physics, andthe Centre for Energy Research, Hungarian Academy ofSciences) have been applied on the three archaeologicalbronze objects.

of Humanities, Hungarian Academy of

gary. E-mail: kiss.viktoria@btk.mta.hu

of Miskolc, 3515 Miskolc, Egyetem Road,

ian Academy of Sciences, 1121 Budapest,

emy of Sciences, 1121 Budapest, Konkoly

ent Istvan Square 26, Hungary

Chemistry 2015

2.1. Neutron radiography (NR)

The guided neutron beams coupled to a research reactor caneffectively be exploited for the radiographic (NR, 2D) andtomographic (NT, 3D) imaging of objects. Due to the differentneutron attenuation coefficients of the sample's inner partsdifferent contrast levels (so-called grey values) can be visualizedin the projected images. These images show then some struc-tural information of the object in a non-destructive way.

The NIPS-NORMA station, a combined elemental analysisand imaging facility, is situated at the end of the neutron guideno. 1 of the Budapest Neutron Centre. The thermal equivalentneutron ux was measured to be 2.7� 107 n cm�2 s�1. The NIPS(Neutron Induced Prompt gamma-ray Spectrometry) facility (seebelow) is used to provide the elemental composition of theobject, in some cases even as spatially resolved information.The NORMA (Neutron Optics and Radiography for Material

Fig. 1 Location of the studied artefacts in Hungary: (1) Zalaszabar,(2) Bonyhad, (3) Abaujdevecser.

J. Anal. At. Spectrom.

Fig. 2 The NIPS-NORMA station of the Budapest Neutron Centre.

JAAS Paper

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

Analysis) facility is used for transmission based neutronradiography and tomography. The components of the station(Fig. 2) could be operated either independently or in acombined mode. The latter serves the experiments yieldingboth composition and inner structural information.

The detailed description of the imaging part (NORMA) ofthe station is as follows. The beam arrives through a 5 � 5 cm2

ight tube into the sample chamber, which has the dimensionsof 20 � 20 � 20 cm3. It is made of an AlMgSi alloy, and linedfrom inside with a neutron absorbing 6Li-enriched polymer. Byremoving one or more side panels, larger objects up to 5 kgweight could be analyzed (such as a sword, vase, stones, etc.).Samples could be positioned to the measurement position in

Table 1 The elemental composition of the flanged axe from Zala-szabar and the axe from Bonyhad given in weight percent (wt%). (<D.L.:under detection limit)

Artefact: spectrumFlanged axe, Zalaszabar639 Axe, Bonyhad 640

Element

Detectionlimit(D.L.) wt% Composition wt% � Composition wt% �

H 0.003 0.056 0.004 0.025 0.003Cu 0.05 90.2 0.7 88.0 0.6Ag 0.05 0.18 0.01 <D.L.Sn 1.0 9.6 0.7 12.0 0.6Zn 0.5 <D.L. <D.L.As 0.2 <D.L. <D.L.

J. Anal. At. Spectrom.

the chamber either manually from the top side, or automaticallyby the use of a XYZ-u sample stage. The stage has a traveldistance of 200 mm with a guaranteed precision of 15 mm.

The NORMA imaging system comprises its main elements ina light-tight house. These are a 100 mm thick, green-light-emitting 6Li/ZnS scintillator, a quartz mirror with the surfacecovered by a thin Al layer, and a cooled, black-and-white, backilluminated Andor iKon M CCD camera with a 16 bit gray valuedepth and 1024 � 1024 pixel resolution. The customized opticsprojects a 48.6� 48.6 mm2 area (in which the beam spot is 40�40 mm2) onto the 13.3 � 13.3 mm2 sensitive surface of the CCDchip. The measured resolution of the imaging system is about230–660 mm depending linearly on the object–screen distance.

The exposure time for the projection is a function of theneutron's energy, and it is 1.8 s for cold neutrons. The acquiredimages are then corrected for both the inhomogeneity of thebeam (at-eld) and the dark current of the camera chip.

Although the neutrons have a large penetration power theirattenuation is heavily dependent on both their energy and thematerial they are traveling. In the case of bronze the coldneutrons can be transmitted through a thickness of some tensof a millimetre. It has a consequence that objects to be sub-jected to tomography should not be larger than these sizesalong the path of the neutron beam.

2.2. Prompt gamma activation analysis (PGAA)

PGAA is a non-destructive and non-invasive nuclear analyticalmethod applicable to determine the elemental composition ofthe samples.1 The samples are irradiated with an external coldneutron beam, and the gamma photons emitted in an (n,g)reaction are detected. PGAA is a panoramic method; in principleall the chemical elements – except helium – can be quantied,without any prior knowledge about the sample.2 Energies andintensities of the gamma lines in the spectra are characteristicof the elemental composition of the samples, but independentof their chemical forms. Since both the incoming neutrons andthe emitted photons are highly penetrating, the provided resultsare volumetric average values characteristic of the whole irra-diated volume.

The PGAA experiments have been performed at the NIPSfacility.3 In the case of the anged axe and the axe, a beam of20 mm2 was used. The acquisition times have been chosen to bebetween 7000 s and 11 500 s, in order to collect statisticallysufficient counts in the spectra.

Detection of gamma photons originated in the (n,g) reactionwas performed in Compton-suppressed mode using a HPGedetector surrounded by BGO scintillators. Signals from thedetector system were collected by a 64k multichannel analyser.Detailed description of the NIPS-NORMA experimental stationcan be found elsewhere.4 Spectrum tting was done withHypermet PC soware.5 Identication of elements was doneusing the ProSpeRo code, which is based on our prompt-gammalibrary.6 The results were corrected for the spectral backgroundmeasured using an open beam without the sample. As a resultof the irradiation with cold neutrons, short-lived radioactivitywas induced which decayed within a few hours.

This journal is © The Royal Society of Chemistry 2015

Paper JAAS

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

The composition of the objects, as measured on the indi-cated parts in Fig. 6 and 7, is given in Table 1. The concentrationdata, their uncertainties, as well as detection limits are given inweight%.

2.3. Time-of-ight neutron diffraction (TOF-ND)

The TOF-ND instrument of BNC is installed at a radial thermalchannel of the reactor. It consists of a 20 m long 25 � 100 mmcross-section neutron guide, a chopper system and detectors.The beam passes through a tunnel from the reactor to themeasuring hall (Fig. 3).

The short neutron-pulses are produced by a fast double-discparallel rotating high speed (12 000 rpm) double chopper. Theshortest pulse length is 10 microseconds, the total ight path ofthe neutrons to the detectors is 25 m. In the highest resolutionmode and backscattering geometry (detector position xed at175�) diffraction spectra with peak widths of 1.5� 10�3 A can bemeasured. The available wavelength range is 0.7–4.5 A, but itcan be covered in one step at lower resolution only. According tothe high penetration depth of the neutrons in copper at theapplied wavelength and the large beam cross-section (10 �2.5 cm), bulk average results can be gained. In the present studythe aim of the diffraction experiment was the phase analysis ofbronze and investigation of its texture.

2.3.1. Phase analysis. In the investigated objects noother alloying elements were observed by PGAA, so the analyseswere conned to the determination of the amounts andconcentration of normal tin-bronze phases, such as a-phaseand g-(probably g-)bronzes. Since no phase other than the solidsolution (a-phase) was found, to avoid the effect of any anisot-ropy, the tin concentration was determined directly from theshape of diffraction peaks. The solute atoms of a solid solutionchange the lattice constants of the host lattice and shi theposition of the diffraction peaks. In the case of a-bronze thischange is proportional to the concentration (Vegard's low). Incasts the concentration distribution is not homogeneous: thechemical concentration within the grains can change from zero

Fig. 3 The time-of-flight neutron diffractometer of the BudapestNeutron Centre.

This journal is © The Royal Society of Chemistry 2015

to the maximal solubility at the temperature of solidication.This inhomogeneity results in peak broadening, and the shapesof the diffraction peaks reect the chemical distribution. Themethod was tested on reference samples, and their concentra-tion was determined to 0.25% precision.7,8 Unless the alloy iscompletely homogenized, the peak is broad, so other propertiessuch as stress, grain size, etc., have no perceptible effect on thepeak shape.

2.3.2. Texture. By the effect of the cold or low temperatureplastic deformation, the originally random orientation ofcrystallites becomes more regular, reecting the type and thedegree of deformation. The preferred orientation or texture isformed, which can fully be described by the orientationdistribution function (ODF). The ODF can be determined froma set of diffraction spectra covering sufficiently large numberof peaks and recorded from sufficiently many directions. Inpractice, the texture generally is characterized by the polegures of several representative diffraction peaks. Since ourdiffractometer is not equipped with an automatic samplerotating mechanism, and the detector solid angle is too small,the texture was analysed by measuring a few spectra at somegiven angles of rotation around two perpendicular axes of theobjects.

2.3.3. Experimental conditions. The instrument was set upto produce neutron-diffraction patterns suitable to determinethe composition and the texture as well. The wavelengthrange was chosen from 1.2 to 2.9 A to cover 8 sufficientlyintense diffraction peaks from (022) to (115). As a compromisebetween the bandwidth, resolution and intensity, moderateresolution (dFWHM¼ 0.03 A) double bandmode was chosen with10 000 rpm parallel rotating pulse choppers. The textures werecharacterised with the relative integrated peak intensities. Theintensities are normalised to a powder-like (fully randomorientation) spectrum. Since the effective scattering volumecannot be determined (because of the irregular shape of theobjects, the back-scattering geometry and the absorption), therelative intensities in one spectrum are determined for acommon factor close to one.

Fig. 4 (1): Flanged axe of the hoard from Zalaszabar, (2): special typebronze axe from Bonyhad.

J. Anal. At. Spectrom.

JAAS Paper

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

3. Investigated artefacts

We investigated three bronze objects from the Middle and LateBronze Age in Hungary (2000–1200 BC; Fig. 1): a anged axe, aspecial type axe, and an oversized, 12 kg arm/ankle ring withspiral ends.

3.1. Flanged axe

The anged axe (L: 12.4 cm) was discovered in a Middle BronzeAge (2000–1500 BC) hoard in Zalaszabar (western Hungary),containing mainly dress ornaments and some tools weighing1.5 kg of bronze (Fig. 4.1).9,10 Large numbers of similar angedaxes11 are oen found in depositions close to the copper minesof western Central Europe; that is why they are usually denedas ingots of copper raw material or special purpose ‘pre-money’.12–14 However, several scholars consider them as toolsbased on post-casting elaboration of their edge.15

The aim of our investigation was to provide data regardingraw material composition and manufacturing techniques, inorder to complete the long discussion about the function of theanged axes, without destruction of the artefacts.

3.2. Special type bronze axe

The special Bonyhad type axe (L: 18 cm; Fig. 4(2)) was excavatedfrom a Middle Bronze Age cremation grave in Bonyhad (westernHungary).16,17 Our investigation targeted the function and usevalue of the special Bonyhad type axe, whether it was used as aprestige weapon or a tool.

3.3. Unique, oversized bronze spiral

The 43 cm large (W: 12 kg) bronze spiral (Fig. 5), dated to theCentral European Middle or Late Bronze Age; (1600–1200 BC),was found in Abaujdevecser-Ortasfoldek (Bekas lake; north-easternHungary) as a single nd.18–22 Based on its size andweightit can be considered as an oversized ritual single nd depot.23 Theaim of our investigation was to analyse the production techniqueand to provide data regarding the raw material.

Fig. 5 Unique, oversized bronze spiral from Abaujdevecser.

J. Anal. At. Spectrom.

4. Results

Using the non-destructive PGAA method, the bulk chemicalcomposition of the tested volumes can be determined. Theresults are shown in Table 1. The exact positions of themeasured volumes within the objects are shown in the guresof the individual subsections (Fig. 6 and 8).

4.1. Flanged axe

4.1.1. Elemental and phase analysis. Based on the resultsof the PGAA method, the determined elemental composition ofthe alloy of the anged axe is 90.2 � 0.7 wt% copper, 9.6 �0.7 wt% tin, 0.18 � 0.01 wt% silver and 0.056 � 0.004 wt%hydrogen. By TOF-ND (A1), the obtained tin content is 7.5 �0.5 wt%. The results of the PGAA and TOF-NDmethod related totin deviate more than one sigma condence interval. A thirddataset from a previous, destructive sampling for energy-dispersive X-ray uorescence (EDXRF) measurement (Table 2)was considered.24 From Table 2 the 8 wt% tin content isconsistent with the TOF-ND within the margin of the uncer-tainty. For the copper and silver content of the anged axe,PGAA and EDXRF results are in good agreement, while someimportant impurities (Fe, Co, Ni, Zn, As, Se, Sb, Te, Au, Pb andBi) were below the detection limits of the of the PGAA.

4.1.2. Production techniques. The radiographic imaging ofthe bronze anged axe was carried out by acquiring 8 over-lapping 2D projections in a tiled manner to be able to cover thewhole object. By the use of the MosaicJ plugin of the FIJI

Fig. 6 (1): The neutron radiography of the flanged axe from Zalasza-bar, (2): PGAA measurement area, (3): area of former destructivesampling (50� 50mm2 shown in panel 1), and the lighter spots aroundit, (4): the 3D surface plot of panel 3.

This journal is © The Royal Society of Chemistry 2015

Fig. 8 Special type bronze axe (code: J37) from Bonyhad.

Table 2 Elemental composition of the flanged axe from Zalaszabar usin

Sample Fe Co Ni Cu Zn As Se

2010.2.1.82 0.02 0.01 0.074 91 0.2 0.129 0.0

Fig. 7 Texture results of the TOF-ND analysis (a): at the neck, (b): atthe edge of the Zalaszabar axe (errors are �0.05–0.01 from thesmallest to the largest d-spacing values).

This journal is © The Royal Society of Chemistry 2015

Paper JAAS

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

image processing soware the tiles can be stitched together toobtain a transmission image of the whole axe (Fig. 6(1)). Thechange in gray values corresponds generally to the differentthicknesses of a homogeneous material. Macroscopic struc-tural deviation can only be found at two distinct areas. Therst is a ssure at the edge of the axe, which can be seen as amagnied area in Fig. 6(2). The small circle with a diameter of5 mm indicates the irradiated area for the purpose of PGAAelemental analysis, and its center is 12 mm away from theedge of the axe. The second area contains several lighter smallspots in the middle part of the axe (Fig. 6(3) and (4)), whichmight correspond to either former samplings, or defects dueto shrink holes or corrosion. Therefore the lighter spots in theradiographic images only indicate presumable surface devi-ations apart from the deviations from former sampling forEDXRF measurements. The differences in the gray values ofthe radiographic image are well recognizable in the 3Dsurface plot of the projection (Fig. 6(4)). The peaks indicatethe lighter spots, and a higher peak means less neutronattenuation. It is of less probability that the spots are due tobubbles created during a casting in a lying position.25 Avail-able projections, however, do not allow for a precise deter-mination of their nature: whether these are sites of corrosionor inhomogeneities caused by casting defects. 3D tomographyinvestigation is suggested in the future to explore the spatial(surface or bulk) position of the mentioned porosity spots.

TOF-ND texture analysis was carried out in two parts of theaxe (Fig. 7):

(1) At the neck part (“ny”), placing it vertically to the beam –

i.e. rotation about the horizontal axis of the object. The angle ofthe scattering vector is measured from the normal direction ofthe at surface. A small part of the head was illuminated as well.

(2) At a 20 mm bandwidth of the edge (“e20”): the axe placedhorizontally – the edge vertically – in the beam, i.e. rotationaround the edge. The angle is measured from the normaldirection of the plane in which the edge lies. Two sets of spectrawere taken from this part.

The spectrum, measured from the perpendicular direction tothe at plane of the neck, indicates isotropic oriental distribu-tion. This suggests that the object is a cast. At higher anglesweak, but signicant, anisotropy appears. Since, at these posi-tions the shoulder and the side of the neck turnedmore into thebeam, the anisotropy can belong to these parts. In this case, theupper layer of the side has to be strongly anisotropic. The type ofthis texture (decreasing intensities in the directions of 100 and111) could be formed by forging of the sides.

The edge is denitely oriented. Turning to the direction of theedge, the intensity of the 100 peak strongly decreases, althoughthe texture can be more complex. It makes it obvious that theedge had been manufactured at least for hardening.

g the EDXRF method (made by Ernst Pernicka)

Ag Sn Sb Te Au Pb Bi

05 0.175 8 0.127 0.005 0.023 0.028 0.01

J. Anal. At. Spectrom.

Fig. 9 Texture results of the TOF-ND analysis (a): at the neck, (b): atthe edge of the Bonyhad axe (errors: �0.07–0.015).

JAAS Paper

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

4.2. Special type axe

4.2.1. Elemental and phase analysis. The PGAA results ofthe special type axe are shown in Table 1. The exact irradiatedvolume is marked in Fig. 9 (round-shape neutron collimator).The centre of the measured area is 18 mm from the edge of theaxe, and the diameter is 5 mm. The bulk composition of theirradiated volume is 88.0 � 0.6 wt% copper, 12.0 � 0.6 wt% tin,and 0.025 � 0.003 wt% hydrogen. The TOF-ND (A2) result forthe tin content is 11.07 � 0.5 wt%.

4.2.2. Production techniques. Due to the difficulties ofpositioning only one NR projection was acquired for the axeobject (code: J37; Fig. 8). The grey scale values could correspondto portions with different thicknesses of a homogeneousmaterial. Macroscopic casting defects (shrink holes) or otherstructural differences are not visible.

TOF-ND texture analysis was carried out in three parts of theaxe:

(1) At the central part (“back”), opposite to the protrusionsvertically (“vert”),

(2) at the edge of the axe (“nail”),(3) at the protrusion parts, in two different setups (“but”).The bulk shows isotropic orientation distribution indicating

that the object was made by casting. Because of the smallvolume of the edge, the spectra taken from here are rathernoisy. Even so, the presence of the texture is demonstrable,although its magnitude is uncertain. The most difficult was theinvestigation of the protrusions. Measuring from the directionperpendicular to its surface, the scattering from themuch largervolume barrel oppresses the signal from the surface layer andseems to be textureless. To separate the upper layer, spectrawere taken from the direction nearly parallel to the surfaces,covering the barrels with the absorber. The measured peakintensities indicate the texture, but the angle dependence israther weak. In summary, we concluded that the shoulders wereformed aer casting by hammering.

Fig. 10 Texture results of the TOF-ND analysis at the round cross-section central part of the Abaujdevecser bronze spiral (errors:�0.02–0.004).

4.3. Oversized bronze spiral

Due to its great size, radiography and PGAA analyses were notacquired for the bronze spiral.

4.3.1. Elemental and phase analysis. By TOF-ND, theobtained tin content of the bronze spiral (code: S) is 10.38 � 0.5wt%.

4.3.2. Production techniques. TOF-ND texture analysis wascarried out in four parts of the spiral (Fig. 10):

(1) In the middle region of one of the coiled up spiral ends(“1a”), from the opposite,

(2) in the middle region of the same coiled up spiral ends(“1b”), from the back,

(3) the octagonal cross-section part of the spiral rod (“2o”),(4) the round cross-section part of the spiral rod (“3o”).Since all the spectra – apart from one – show isotropic orien-

tation distribution, indicating that the object is cast, thedescription of every single experiment would bemeaningless. Theonly exception is the rst one, where at 40 degree a weak anisot-ropy (hardly beyond the error) can be observed. It could come to

J. Anal. At. Spectrom.

existence during casting or nishing the surface, or some largercrystallites are present at this part.

5. Conclusions5.1. Elemental and phase analysis

In the case of the anged axe from Zalaszabar, it was alsopossible to compare different analytical methods. Non-destructive TOF-ND analysis yielded results similar to theEDXRF method (Table 2), while the PGAA analysis providedsomewhat discordant results regarding tin concentrations(Tables 1 and 2) that can be explained by the non-homogeneouscomposition of the prehistoric bronze objects. The PGAA andTOF-ND tin results of the special Bonyhad type axe agreedwithin uncertainty limits.

This journal is © The Royal Society of Chemistry 2015

Paper JAAS

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

EDXRF analysis of the anged axe from Zalaszabar showedelevated arsenic, silver and antimony contents, suggesting thatthe artefact was produced using an ore obtained from a depositdominated by fahlores and chalcopyrite.26 Silver content couldbe well measured by PGAA, but arsenic and antimony werebelow the detection limit.

5.2. Production techniques

The examined anged axe (Fig. 4.1) was worked aer casting atleast at the edges. Metallographic studies of numerous angedaxes are in concordance with these results: at the neck the metalstructure suggests nothing else, but an as-cast microstructure.Near the edge, however, an ultrastructure, typical of working,could be observed.27 A signicant advantage of TOF-ND analysiswas that we could reproduce the same results with a non-destructive method.

The neutron radiographic image showed light areas atdifferent locations that can be interpreted as presumablesurface deviations, probably sites of corrosion, apart from thedeviations from former sampling for EDXRF measurements. Acasting in a standing position would be supported by thepresence of bubbles at the upper end of the axe, but suchbubbles could not be found in the case of this axe from Zalas-zabar. Based on other technical details of production, theposition of the fracture surface of the broken casting sprue, theaxe was cast in a standing position. The homogeneity of theupper end of the axe (Fig. 6(1)), supported by the lack ofbubbles, shows a good quality casting procedure. However,more exact localization of porosity spots would be possible by3D tomography with higher energy, thermal neutrons.

The special Bonyhad type axe (Fig. 4(2)) was cast and thenworked at the edge (hardened and sharpened by hammering)which also suggests that it was used as a tool. The two smallprotrusions that also helped to x the axe to the handle wereoriginally casting gates and their surface was later smoothenedby hammering. This hammering resulted in the formation ofanges on them. The hammered surface was then decorated bypunctuation (Fig. 4(2)). Macroscopic casting defects (shrinkholes) or other structural differences are not visible in theneutron radiographic image (Fig. 8).

Analysis of the bronze spiral from Abaujdevecser (Fig. 5)conrms that the lost-wax casting method was used during itsproduction. First, a long wax rod was made, with an octagonalcross-section of its middle region. The thinner end parts werecoiled up, mimicking the typical design of smaller, hammeredarm/ankle rings with spiral ends. This resulted in mildly conicaldisks that were smoothened on the upper surface, keeping thestreamline of cast metal also in mind. This process can be wellreconstructed from the traces of coiling, differing on the twosurfaces and the ne, crack-like lines created by the surface ofthe wax. Aer shaping the mass of the artefact, small pieces ofwax were attached to the model at the centers of the disks, inorder to place casting gates. This can be very well observed onone of the disks (Fig. 5), where the dimensions of the removedsprue can also be measured on the imperfectly hammered,otherwise typically granular surface. Before placing it into the

This journal is © The Royal Society of Chemistry 2015

casting mold, smaller corrections were also made on the arte-fact; one end was cut away, for example and a slant surface wasestablished. During the casting process, the mold was placedhorizontally, with the conical disks facing upwards (Fig. 5).

6. Summary

Regarding the elemental composition, it is important tocompare the results of multiple methods (non-destructive PGAAand TOF-ND as well as EDXRF) when analyzing historical arte-facts of non-homogeneous compositions. EDXRF is bettersuited for analysis of the surface layers, containing higherquantities of tin,28 while mainly bulk raw material compositiondata could be obtained from PGAA and TOF-ND analysis.

NR and TOF-ND study of the three artefacts has providednew information regarding the different phases of productionand shed light on the function of the artefacts. Forging, assuggested by TOF-ND results, provides important informationon the function of the special type axe from Bonyhad that couldotherwise, based on its extraordinary shape, be interpreted as asymbolic sign of prestige. Additionally, in the case of theanged axe from Zalaszabar, these results also help in makingthe decision whether this artefact should be regarded as aningot or a tool.29,30 Forged edges of both axes provide strongevidence supporting that these artefacts were indeed used astools. Our analyses show an anisotropic structure proving post-casting elaboration (recrystallization), due to forging andannealing carried out following the casting procedure, on theedges of the axe and the anged axe. It is important to note thatwhile post-casting elaboration was formerly identied bydestructive microstructure analysis of anged axes,31 we coulddetect the same without sampling or any destruction of theartefacts. Neutron radiographic images showed some sign ofinhomogeneity, indicating signs of corrosion rather thanimperfect casting.

The completely isotropic structure of the arm/ankle ringfrom Abaujdevecser, according to TOF-ND results, supportsthe technological observations that the object was made by thelost-wax casting technique without any other post-castingprocedures.

Based on the above we provide data completing the longdiscussion about the function of Early Bronze Age raw materialingots with the analysis of a anged axe, without destruction ofthe artefacts. Forging suggests that the axes were used as tools,and not as prestige weapons or ingots, however anged axescould also serve as ingots.32 A signicant advantage of ouranalysis was that we could produce results on productiontechniques with non-destructive methods that are very impor-tant for the investigation of unique prehistoric metal artefactsconsidering aspects of heritage protection.

Acknowledgements

EDXRF analysis was carried out at the Tubingen University in2006, during the project Untersuchungen zur Vermittlung derZinnbronze nach Mitteleuropa uber das Karpatenbecken, led byErnst Pernicka and Tobias L. Kienlin. We would like to thank for

J. Anal. At. Spectrom.

JAAS Paper

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

the opportunity to publish the data of the EDXRF result of theanged axe.

References

1 Z. Revay and T. Belgya, Principles of the PGAA method, in:Handbook of Prompt Gamma Activation Analysis with NeutronBeams, ed. G. L. Molnar, Kluwer Academic Publishers,Dordrecht/Boston/New York, 2004, pp. 1–30.

2 Z. Revay, Determining Elemental Composition Using ProptGamma Activation Analysis, Anal. Chem., 2009, 81, 6851.

3 L. Szentmiklosi, Z. Kis, T. Belgya and A. N. Berlizov, On thedesign and installation of a Compton-suppressed HPGespectrometer at the Budapest neutron-induced promptgamma spectroscopy (NIPS) facility, J. Radioanal. Nucl.Chem., 2013, 298, 1605–1611.

4 L. Szentmiklosi, T. Belgya, Zs. Revay and Z. Kis, Upgrade ofthe prompt gamma activation analysis and the neutron-induced prompt gamma spectroscopy facilities at theBudapest research reactor, J. Radioanal. Nucl. Chem., 2010,286, 501–505.

5 Zs. Revay, T. Belgya and G. L. Molnar, Application ofHypermet-PC in PGAA, J. Radioanal. Nucl. Chem., 2005, 265,261–265.

6 Z. Revay, R. B. Firestone, T. Belgya and G. L. Molnar, Catalogand Atlas of Prompt Gamma Rays, in Handbook of PromptGamma Activation Analysis with Neutron Beams, ed. G. L.Molnar, Kluwer Academic Publishers, Dordrecht/Boston/New York, 2004, pp. 173–364.

7 M. Modlinger, Zs. Kasztovszky, Z. Kis, B. Maroti, I. Kovacs,Z. Sz}okefalvi-Nagy, Gy. Kali, E. Horvath, Zs. Santa and Z. ElMorr, Non-invasive PGAA, PIXE and ToF-ND analyses onHungarian Bronze Age defensive armour, J. Radioanal.Nucl. Chem., 2014, 300, 787–799.

8 M. Modlinger, P. Piccardo, Zs. Kasztovszky, I. Kovacs,Z. Sz}okefalvi-Nagy, Gy. Kali and V. Szilagyi,Archaeometallurgical characterization of the earliestEuropean metal helmets, Mater. Charact., 2013, 79, 22–36.

9 V. Kiss, The Life Cycle of Middle Bronze Age Bronze Artefactsfrom theWestern Part of the Carpathian Basin, inMetals andSocieties, Studies in Honour of Barbara S. Ottaway, ed. T. L.Kienlin and B. Roberts, Universitatsforschungen zurprahistorischen Archaologie 169, Bonn, 2009, pp. 328–335.

10 S. Honti and V. Kiss, Bronze Hoard from Zalaszabar. NewData on the Study of the Tolnanemedi Horizon – Part 2, inMoments in Time: Papers Presented to Pal Raczky on His 60thBirthday. Prehistoric Studies I, ed. A. Anders, G. Kulcsar, G.Kalla, V. Kiss and G. V. Szabo, Budapest, 2013, pp. 521–538.

11 T. L. Kienlin, Traditions and transformations: approaches toEneolithic (Copper Age) and Bronze Age metalworking andsociety in eastern Central Europe and the CarpathianBasin, BAR International Series 2184, Oxford, 2010, Fig. 7.2.

12 M. Lenerz-de Wilde, Pramonetare Zahlungsmittel in derKupfer- und Bronzezeit Mitteleuropas, Fundberichte ausBaden-Wurttenberg, 1995, 20, pp. 229–327.

13 J. Muller, Modelle zur Einfuhrung derZinnbronzetechnologie und zur sozialen Differenzierung

J. Anal. At. Spectrom.

der mitteleuropaischen Fruhbronzezeit, in: VomEndneolithikum zur Fruhbronzezeit: Muster sozialenWandels?, Universitatsforschung zur prahistorischenArchaologie 90, ed. J. Muller (Hrsg.), Bonn 2002, pp. 267–289.

14 R. Krause, Studien zur kupfer- und fruhbronzezeitlichenMetallurgie zwischen Karpatenbecken und Ostsee,Vorgeschichtliche Forschungen 24, Rahden/Westfalen, 2003.

15 T. L. Kienlin, Traditions and transformations: approaches toEneolithic (Copper age) and Bronze Age metalworking andsociety in eastern Central Europe and the Carpathianbasin, BAR International Series 2184, Oxford, 2010.

16 G. Szabo, Csalog Jozsef bonyhadi meszbetetes temet}oje azujabb kutatasi eredmenyek es ket egyedulallo lelettukreben – The cemetery of the Encrusted Pottery cultureat Bonyhad in the light of new research and two uniquends, in Medinatol Eteig. Tisztelg}o Irasok Csalog JozsefSzuletesenek 100. Evfordulojan, ed. L. Bende and G.L}orinczy, Szentes, 2009, pp. 283–291.

17 G. Szabo, El}omunkalatok Bonyhad, Pannonia Zrt. Biogaz UzemMegel}oz}o Regeszeti Feltaras Anyaga Feldolgozasahoz I(Sırmellekletek, Embertan), DVD, Szekszard, 2012.

18 M. Hellebrandt, Bronztekercs Abaujdevecserr}ol – Bronzespiral of Abaujdevecser, Herman Otto Muzeum Evkonyve,2011, 50, pp. 131–152.

19 T. Kovacs, Die Bestattungsriten der Fuzesabony-Kultur unddas Graberfeld von Tiszafured-Majoroshalom, in Bronzezeitin Ungarn. Forschungen in Tell-Siedlungen an Donau undTheiss. Frankfurt a.M, ed. W. Meier-Arendt (Hrsg.), 1992,pp. 96–98, Abb. 62.

20 J. Koos, Der II. Bronzefund von Tiszaladany,Communicationes Archaeologicae Hungariae, 1989, pp. 31–43.

21 W. David, Studien zu Ornamentik und Datierung derbronzezeitlichen Depotfundgruppe Hajdusamson-Apa-Ighiel-Zajta, Bibliotheca Musei Apulenis 18, Alba Iulia, 2002.

22 G. Szabo, What Archaeometallurgy Tells Us about theChanges of Bronze Crawork in the Carpathian Basin atthe Transition of the Bronze Age into Iron Age, in BronzeAge Cras and Crasmen in the Carpathian Basin.Proceedings of the International Colloquium from TarguMures 5–7 October 2012. Bibliotheca Mvsei Marisiensis, SeriaArchaeologica VI, ed. B. Rezi, E. R. Nemeth and S. Berecki.Targu Mures 2013, pp. 291–312.

23 A. Krenn-Leeb, Ressource versus Ritual –

Deponierungsstrategien der Fruhbronzezeit in Osterreich,in Der Griff nach den Sternen. Internationales Symposium inHalle (Saale) 16.–21. Februar 2005. Tagungen desLandesmuseums fur Vorgeschichte Halle 5, ed. H. Meller, F.Bertemes and F. (Hrsg.), Halle/Saale, 2010, pp. 281–315.

24 J. Lutz and E. Pernicka, Energy dispersive X-ray analysis ofancient copper alloys: empirical values for precision andaccuracy, Archaeometry, 1996, 38, 313–323.

25 T. L. Kienlin, Traditions and Transformations: Approaches toEneolithic (Copper Age) and Bronze Age Metalworking andSociety in Eastern Central Europe and the Carpathian Basin,BAR International Series 2184, Oxford, 2010, Fig. 7.13.

This journal is © The Royal Society of Chemistry 2015

Paper JAAS

Publ

ishe

d on

15

Dec

embe

r 20

14. D

ownl

oade

d by

RSC

Int

erna

l on

16/0

1/20

15 1

3:06

:07.

View Article Online

26 V. Kiss, P. Barkoczy and Zs. Vızer, Preliminary Results of theMetallographic Analysis of the Zalaszabar Hoard, Gesta,2013, 13.

27 T. L. Kienlin, Traditions and Transformations: Approaches toEneolithic (Copper Age) and Bronze Age Metalworking andSociety in Eastern Central Europe and the Carpathian Basin,BAR International Series 2184, Oxford, 2010, Fig. 7.19.

28 G. SzaboAz archaeometallurgiai kutatasok gyakorlati esetikai kerdesei – Practical and ethical issues ofarchaeometallurgical research, Archeometriai M}uhely, 2010,7/2, pp. 111–122.

29 G. Szabo, The manufacture and usage of Late Bronze Agerings: two new ring hoards, in Studien zur Metallindustrieim Karpatenbecken und den benachbarten Regionen,Festschri fur Amalia Mozsolics zum 85. Geburtstag, ed. T.Kovacs, Budapest, 1996, pp. 207–230.

This journal is © The Royal Society of Chemistry 2015

30 V. Kiss, The Life Cycle of Middle Bronze Age Bronze Artefactsfrom theWestern Part of the Carpathian Basin, inMetals andSocieties. Studies in Honour of Barbara S. Ottaway, ed. T. L.Kienlin and B. Roberts, Universitatsforschungen zurprahistorischen Archaologie 169. Bonn, 2009, pp. 328–335.

31 T. L. Kienlin, Fruhbronzezeitliche Randleistenbeile vonBohringen-Rickelshausen und Hindelwangen: Ergebnisseeiner metallographischen Untersuchung, PrahistorischeZeitschri, 2006, 81, 97–120.

32 H. Vandkilde, A Biographical Perspective on Osenringe fromthe Early Bronze Age, in Die Dinge als Zeichen: KulturellesWissen und materieller Kultur, Internationale Fachtagungan der Johan Wolfgang Goethe-Universitat, Frankfurt amMain 3.-5. April 2003, ed. T. L. Kienlin (Hrsg.),Universitatsforschungen zur prahistorischen Archaologie125, Bonn, 2005, pp. 263–281.

J. Anal. At. Spectrom.