Download - Mössbauer Study of Ceramic Finds from the Galería de las Ofrendas, Chavín de Huántar

Hyperfine Interactions 150: 51–72, 2003.© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

51

Mössbauer Study of Ceramic Finds from theGalería de las Ofrendas, Chavín de Huántar

L. G. LUMBRERAS1, R. GEBHARD2, W. HÄUSLER3,F. KAUFFMANN-DOIG4, J. RIEDERER5, G. SIEBEN3 and U. WAGNER3

1Museo Nacional de Antropología, Pueblo Libre, Lima 21, Peru2Archäologische Staatssammlung München, Lerchenfeldstraße 2, 80538 München, Germany3Physik-Department E15, Technische Universität München, 85747 Garching, Germany4Universidad Peruana de Ciéncias Aplicadas (UPC), Lima, Peru5Rathgen-Forschungslabor, Schloss-Str. 1a, 14059 Berlin, Germany

Abstract. Ceramic finds from the Galería de las Ofrendas at Chavín de Huántar and surface findsfrom the settlement of Chavín were characterised by combining the results of archaeological typol-ogy with archaeometric studies using neutron activation analysis, Mössbauer spectroscopy, X-raydiffraction and thin-section microscopy. Sherds from the pyramid Tello are included in the study asrepresentative of local material. The analyses show that the vessels were made from different rawmaterials and that different firing procedures were used in their production. Sherds of certain styleslargely exhibit similar types of Mössbauer patterns and in many instances also have similar elementcompositions. This supports the archaeological notion that the vessels were brought to Chavín fromthe provinces, perhaps on the occasion of a festivity.

Key words: Formative ceramics, Galería de las Ofrendas, Chavín de Huántar, Mössbauer spec-troscopy, thin section microscopy, X-ray diffraction.

1. Introduction

Chavín de Huántar is a Pre-Columbian ceremonial site in the Andes of northernPeru, at an altitude of 3180 m above sea level at the foot of the Cordillera Blanca.In the precinct of the main temple at Chavín de Huántar, a number of undergroundgalleries were found. During the years 1966 and 1967, when one of these wasexcavated by Lumbreras, about eighteen thousand ceramic sherds were recovered,which could be shown to stem from nearly 700 vessels [1, 2]. Since these wereconsidered as offerings brought to the sanctuary by pilgrims, the gallery was calledthe Galería de las Ofrendas. Although the Galería de las Ofrendas represents aclosed archaeological context, the styles of the ceramics found there differ widelyfrom each other and from those of local ware, which seems to support the notionof offerings brought to Chavín from the surrounding provinces, perhaps on theoccasion of a single festivity. We have studied 70 specimens of ceramic finds fromthe Galería de las Ofrendas by Mössbauer spectroscopy, X-ray diffraction, thin

52 L. G. LUMBRERAS ET AL.

section microscopy and neutron activation analysis. Data from previous studies ofsherds from the Pyramid Tello at Chavín de Huántar [3] and from the settlementof Chavín [4–6] are referred to for comparison. The rationale behind this inves-tigation is to learn if vessels of different archaeological styles also have differentmaterial properties. This would confirm the notion that they were made of clayfrom different locations and brought from there to Chavín, rather than being madelocally.

The archaeological classification divides the samples from the Galería de lasOfrendas into different styles with names such as Wacheqsa, Floral or Puksha [1].In order to uniquely describe the individual sherds studied in this work, they wereadditionally given numbers in the database we maintain at our Garching laboratory(GarNum, e.g., 13/73). In the following, the sherds will be designated by theseGarNums and, in addition, by the archaeological style name.

2. Neutron activation analysis

Neutron activation analysis (NAA) [7] was used to determine the concentrationsof 18 trace and 3 major elements in 64 finds from the Galería de las Ofrendas.Previously 53 surface finds from the settlement of Chavín and 61 sherds from thePyramid Tello were studied [4–6] and can be used together with the present data.The dendrogram obtained by a cluster analysis of the element concentrations forthe finds from the Galería using the average weighted linkage method (Figure 1,left) has been interpreted in terms of eight groups, though any such subdivision isto some extent arbitrary. However, notwithstanding the chosen partition, ceramicsof certain archaeological styles should be found close together in the dendrogramif they came from the same place or even workshop.

The archaeological classification distinguishes [1] between local Chavín ceram-ics and other styles, namely Mosna, Puca Orqo, Puksha, Raku and Wacheqsa. Thelarge class of Chavín ceramics is subdivided into the styles Ofrendas, Dragoniano,Qotopukyo and Floral, which themselves are again subdivided. The non-Chavínstyles are supposed to form close groups, which should be near together in thedendrogram. Indeed the Mosna style sherds form a well-separated, compact cluster(NAA-group 5) containing all studied Mosna specimens but no others. The sixstudied Wacheqsa specimens are all in NAA-group 1 and the 4 studied Pukshaspecimens are all in NAA-group 7, albeit together with a number of other styles.Thus, for Mosna, Wacheqsa and Puksha, there is a high degree of similarity inthe NAA data. This is not the case for the Raku ceramics, where the 5 studiedspecimens are divided between NAA-groups 1 and 3, nor for Puca Orqo, where the8 studied specimens are scattered over NAA-groups 1, 3, and 7.

The various styles of Chavín-type ceramics are widely scattered over the den-drogram, though not completely devoid of some systematic behaviour. Thus, Drag-

CERAMIC FINDS FROM THE GALERIA DE LAS OFRENDAS 53

Figure 1. Cluster analysis of the element concentrations determined by neutron activation analysisin 65 finds from the Galerıa de las Ofrendas (left). The dendrogram yields 8 groups. Surface finds(labelled with an ∗) and 61 sherds from the Pyramid Tello (represented by a single bar) are includedin the dendrogram on the right. There is no similarity between the finds from the Galerıa de lasOfrendas and from the Pyramid Tello. Some of the surface finds, however, cluster together withsherds from the Galerıa.


oniano is only in NAA-group 3, while the Ofrendas rojo sherds form two distinctbut compact NAA-groups, 4 and 6. Floral is in the widely different NAA-groups 1and 7, and Ofrendas gray is widely spread over NAA-group 1, 2, 3, and 7.

When the NAA data for the Galería de las Ofrendas are clustered together withthose for sherds from the Pyramid Tello and the surface finds from the settlementof Chavín (Figure 1, right), the material from the Pyramid Tello, which has beenconsidered as locally made [4–6], definitely forms an independent group. This canbe considered as evidence in favour of the notion that the vessels from the Galeríade las Ofrendas were not made locally at Chavín. Part of the surface finds blend intothe groups formed by the sherds from the Galería (Figure 1, right). This may, insome cases, be incidental, but there is at least one case where mere coincidence canvirtually be ruled out: Ten of the surface finds form a very compact cluster togetherwith a sherd from the Galería de las Ofrendas (NAA-group 8) stemming from avessel of the style called Floral black, of which only this one vessel was found inthe Galería. These 11 specimens also have practically identical Mössbauer spectra(see Section 4.1), which confirms their common origin.

3. Thin-section microscopy

Thin sections were made of 65 sherds and analysed by optical microscopy [8]for texture and mineral content. The dominant minerals are varying amounts ofquartz, plagioclase, hornblende and biotite. Occasionally rock particles, clay nod-ules, hematite and magnetite as well as pores and plant inclusions are observed.Samples of the same archaeological style generally show similar thin sections,although a certain variation is observed in the detail. One sherd of each NAA groupis shown together with the respective thin section in Figures 2 and 3.

The sherds of the Wacheqsa type (NAA-group 1) form a rather homogeneousgroup with inclusions of medium size corresponding to texture C3 and D3 accord-ing to Riederer [8]. Sherds of the type Puca Orqo red (NAA-group 3) contain rockparticles and hematite and are somewhat coarse grained. Sherds of the Qotopukyotype (also NAA-group 3) have rock and plant inclusions as prominent properties.NAA-group 4 contains only two Ofrendas rojo sherds which are, however, strik-ingly different from all the others, containing large amounts of coarse inclusions.Samples of the Mosna type (NAA-group 5) are relatively uniform, with mediumgrain sizes and many plagioclase inclusions. NAA-group 6 contains six similarsherds of the type Ofrendas rojo with coarse-grained inclusions and additionalhematite. In NAA-group 7 four uniform and coarse grained sherds containing rockparticles are found, the other members of this group being rather inhomogeneous.NAA-group 8 consists only of one sherd of Floral black, containing large amountsof hornblende.

The obvious differences in the mineral content of the individual NAA groupsindicates a different provenance of the different groups. In many cases the resultsof the thin section study yield decisive clues for the classification of the sherds.


Figure 2. Characteristic thin-section micrographs (right) of sherds (left) from the Galerıa de lasOfrendas belonging (from top to bottom) to NAA-groups 1 to 4. The dominant minerals are plagio-clase (P), quartz (Q), hornblende (H), biotite (B), rocks with different inclusions (R*), opaque ore(O*) and hematite (He). The scale bar on the left is 1 cm long, on the right it is 1 mm.


Figure 3. Characteristic thin section micrographs (right) of sherds (left) from the Galerıa de lasOfrendas belonging (from top to bottom) to NAA-groups 5 to 8. The dominant minerals are plagio-clase (P), quartz (Q), hornblende (H), opaque ore (O*) and serecite (S). The scale bar on the left is1 cm long, on the right it is 1 mm.


4. Mössbauer spectroscopy and X-ray diffraction

4.1. THE TYPES OF MÖSSBAUER SPECTRA FOUND IN CERAMICS FROM

CHAVÍN

The Mössbauer spectra of ceramics reflect the chemical and physical state of theiron after firing. X-ray diffraction, on the other hand, yields a status of all minerals,but the iron-containing ones often have too small concentrations to be seen. Thetwo methods therefore complement each other. The mineral content of ceramicsdepends on both the raw materials and the firing procedure. The raw materials maybear information on the provenance. The firing techniques are often characteristicfor certain workshops and therefore also bear implicit information on the prove-nance of ceramics. What we call the firing technique involves many parameters,like the firing temperature, the duration of the firing and the kiln atmosphere. Mat-ters are further complicated by variations of the temperature and kiln atmospherein the course of the firing process. All firing parameters may have been adapted bythe ancient potters to achieve the desired properties and looks of the vessels. Oftenthe surfaces of the vessels were polished or decorated by the application of a slip.Such surface treatments can be studied to depths of several hundred nanometres byconversion electron Mössbauer spectroscopy (CEMS) and to depths of several tensof µm by Mössbauer spectroscopy with backscattered gamma rays or X-rays [9].Such experiments were performed in a few cases on ceramics from the Galería delas Ofrendas, though most of the work was done by standard absorption Mössbauerspectroscopy.

Adsorption Mössbauer spectra were measured at room temperature (RT) and inselected cases also at the temperature of liquid helium (4.2 K) with a source of 57Coin a rhodium matrix kept at the same temperature as the absorber. A surface layerof about a millimetre thickness of the sherds is often visibly different from the coreof the sherd. In such cases the interior is usually gray, while the surface is red orbrown, which already indicates that the iron in the surface layer is mainly trivalent,while it is mainly divalent in the cores. In such cases material from the surfaceand the core of the sherds was studied separately. Despite a certain variability ofthe Mössbauer spectra, dominant features could be established, which allowed thesamples to be classified according to the type of their Mössbauer patterns. Thedifferent types of Mössbauer spectra (Mos-types) were defined on account of thepresence and intensity or the absence of certain components, mainly in the RTMössbauer spectra. The use of eight different Mos-types, named A through H, is tosome extent arbitrary but appeared to be a viable compromise between a detailedcharacterisation and a limited number of different types. Only material from thecores was used for the characterisation of the Mos-types in order to avoid anyinfluence of surface treatments. Typical Mössbauer spectra of the eight differentMos-types are shown in Figures 4 and 5. Typical Mössbauer parameters for theindividual Mos-types are compiled in Table I. X-ray diffraction patterns of the


Figure 4. Characteristic Mössbauer spectra of Mos-types A through D for the cores of sherds foundin the Galerıa de las Ofrendas, Chavın. Measurements at RT (left) and 4.2 K (right) are shown.

samples whose Mössbauer spectra are shown in Figures 4 and 5 are reproduced inFigures 6 and 7.

For Mos-types A through D the RT spectra are dominated by an Fe3+ quadru-pole doublet, while in the 4.2 K spectra the sextet of hematite and a broad magnet-ically split background (fitted with a distribution of hyperfine fields) prevail. This


Figure 5. Characteristic Mössbauer spectra of Mos-types E through H for the cores of sherds foundin the Galerıa de las Ofrendas, Chavın. Measurements at RT (left) and 4.2 K (right) are shown. Notethe different velocity scales.


Table I. Characteristic Mössbauer parameters for the cores of 8 selected sherds measured at RTand 4.2 K. The Mos-types A through H were defined mainly on the basis of the Mössbauer spectra(see text). B is the magnetic hyperfine field, Q the quadrupole splitting and IS the isomer shiftwith respect to the source of 57Co in Rhodium [9]. A-mag is the fractional area of the magneticallysplit component(s), A that of the quadrupole doublets. The magnetic components are hematite (hem,magnetite (magn), a distribution of magnetic fields (dis) and an octet for magnetically split Fe2+(oct)

Mos Sample & Mag B A-mag Species Q IS A

type temp. species Tesla % mm/s mm/s %

A 13/12 hem. 51.3 4.2 Fe3+(1) 0.79 0.28 57.9

Wacheqsa magn. 45.4 6.7 Fe3+(2) 1.63 0.39 16.5

RT magn. 48.5 4.6 Fe2+(1) 2.93 0.75 10.1

A 13/12 hem 52.6 32.0 Fe3+(1) 1.06 0.24 22.8

Wacheqsa dis 10–49 34.9 Fe2+(1) 2.39 1.22 10.3

4.2 K

B 13/20 hem 51.0 6.4 Fe3+(1) 0.88 0.28 56.7

Puksha magn. 45.8 13.4 Fe2+(1) 2.20 0.94 11.5

RT magn. 48.8 7.6 Fe2+(2) 2.68 0.87 4.4

B 13/20 hem 52.6 34.7 Fe3+(1) 1.01 0.25 7.4

Puksha dis 6–49 43.9 Fe2+(1) 2.39 1.07 14.0

4.2 K

C 13/73 hem 51.2 4.8 Fe3+(1) 1.07 0.27 49.5

Floral magn. 45.7 9.5 Fe3+(2) 2.04 0.25 11.1

RT magn. 48.7 6.2 Fe2+(1) 2.36 1.04 18.9

C 13/73 hem 52.9 18.4 Fe3+(1) 0.80 0.34 9.4

Floral hem(2) 50.8 4.0 Fe2+(1) 2.24 0.69 3.1

4.2 K dis 6–49 49.2 Fe2+(2) 2.80 1.01 9.8

oct 12.0 6.1

D 13/2 hem 50.4 43.3 Fe3+(1) 0.90 0.21 56.7

Puca Orqo red

RT

D 13/2 hem(1) 53.0 19.0 Fe3+(1) 1.01 0.18 6.9

Puca Orqo hem(2) 52.8 24.5

4.2 K dis 6–49 49.6

E 13/33 hem 48.9 7.7 Fe3+(1) 0.63 0.44 23.6

Raku Fe3+(2) 1.83 0.36 5.8

RT Fe2+(1) 1.88 0.90 30.5

Fe2+(2) 2.45 0.90 14.9

Fe2+(3) 2.51 1.12 17.5


Table I. (Continued)

Mos Sample & Mag B A-mag Species Q IS A

type temp. species Tesla % mm/s mm/s %

E 13/33 hem 52.2 12.1 Fe2+(1) 2.04 0.98 22.2

Raku dis 7–49 29.0 Fe2+(2) 2.73 1.08 19.9

4.2 K oct 12.0 16.8

F 13/28 hem 49.4 8.8 Fe3+(1) 0.77 0.36 32.8

Mosna dis 3–45 11.3 Fe2+(1) 1.88 0.84 16.7

RT Fe2+(2) 2.21 0.96 19.3

Fe2+(3) 2.70 1.01 11.1

F 13/28 hem 52.4 20.9 Fe2+(1) 2.05 0.90 27.2

Mosna dis 4–49 23.4 Fe2+(1) 2.54 1.10 16.5

4.2 K oct 14.0 12.0

G 13/61 hem 46.7 12.3 Fe3+(1) 0.98 0.17 26.7

Ofr.Rojo Fe3+(2) 1.07 0.44 20.0

RT Fe2+(1) 1.47 0.64 8.0

Fe2+(2) 2.58 0.71 21.9

Fe2+(3) 2.25 1.22 11.1

G 13/61 hem 52.5 12.6 Fe3+(1) 0.87 0.30 10.5

Ofr.Rojo hem(2) 51.2 3.8 Fe2+(1) 2.49 0.75 5.6

4.2 K dis 8–49 45.3 Fe2+(2) 2.63 1.11 19.0

oct 12.0 3.2

H 13/75 Fe3+(1) 0.95 0.19 30.9

Floral Fe3+(2) 1.64 0.09 8.0

black Fe2+(1) 1.06 0.69 19.4

Fe2+(2) 2.17 0.91 21.6

Fe2+(3) 2.32 1.12 7.7

Fe2+(4) 2.87 1.02 12.4

H 13/75 hem 52.8 7.5 Fe3+(1) 0.77 0.39 11.1

Floral bl. dis 7–49 40.4 Fe2+(1) 2.46 0.73 4.8

4.2 K oct 12.0 13.0 Fe2+(2) 2.84 1.08 23.2

background is attributed to Fe3+ in the matrix of the dehydroxilated clay minerals,and may be caused by slow paramagnetic relaxation if the iron spins are dilute, orto spin-glass magnetic order if the iron concentration is higher [10]. Fe2+ is alwaysweak in Mos-types A through D. It is strongest in type C with 19% and almostalways absent in type D. This shows that the firing conditions were oxidising atleast at the end of the firing cycle. Mos-types E through H have RT spectra inwhich the quadrupole doublets of Fe2+ dominate, most strongly in type E with


Figure 6. Characteristic XRD patterns for Mos-types A through D. The peaks are labelled with therespective mineral names.

73%, and least in type F with 47%. Magnetically split hematite is always weakat RT. The 4.2 K spectra are dominated by magnetically split components similar tothose also found in types A through D, while the Fe2+ quadrupole doublets havedecreased to less than half of their relative intensities in the RT spectra except intype F, where the decrease is small. This decrease shows that part of the Fe2+ splits


Figure 7. Characteristic XRD patterns for Mos-types E through H. The peaks are labelled with therespective mineral names.

magnetically at 4.2 K and then blends into the dominant magnetically split Fe3+components, which are always showing a substantial broad background in whichthe Fe2+ is difficult to locate reliably. Efforts were made to take the magneticallysplit Fe2+ into account by fitting a magnetic octet component to the 4.2 K spectra,but the relative intensities and hyperfine fields for these can only be considered


as tentative. In any case, the strong presence of Fe2+ shows that Mos-types Ethrough H were subject to reducing firing conditions at least towards the end ofthe firing cycle.

In addition to the distinctive Fe2+ and Fe3+ contents, the individual Mos-typesexhibit other typical features by which they can be further differentiated. Thus,the RT Mössbauer spectra of Mos-types A, B and C contain a small magnetitecomponent, which cannot be distinguished from that of hematite and the broadmagnetic background at 4.2 K. The presence of magnetite was, however, confirmedby XRD on samples separated from the powdered ceramics with a magnet. Thesimultaneous presence of both magnetite and hematite indicates a non-equilibriumstate between oxidising and reducing firing conditions that may have arisen from achanching kiln atmosphere.

The RT Mössbauer spectra of Mos-type A exhibit an additional Fe3+ doubletwith a comparatively large splitting of 1.6 mm/s, which may be due to the presenceof biotite [11]. X-ray diffraction cannot unambiguously distinguish between biotiteand muscovite, due to the low concentration of the mica minerals and the highquartz content, because the 211 peak of quartz is superimposed on the diagnostic060 peak of biotite [12]. Apart from this, the main difference between Mos-types Aand B is the intensity of the Fe2+ component, which is stronger in type B. The X-raydiffractograms show no decisive difference between Mos-types A and B.

Mos-type C spectra, represented by sherd 13/73 in Figure 5, contain a pro-nounced Fe3+ component with an unusually large quadrupole splitting of2.04 mm/s, which is no longer discernible at 4.2 K, where it seems to be magnet-ically split and hidden in the broad magnetic background. This is typical [13–15],for epidote, Ca2(Al,Fe)3Si3O12OH. The XRD pattern of 13/73 exhibits a strongamphibole peak at 12◦2� and several peaks in the region of 5–12◦2� due tothe presence of different layer silicates. A fit of the diffractogram in the regionof 5–25◦2� (Figure 8, top) shows that these peaks are due to chlorite, mixed-layer silicates, mica and steatite [16]. The presence of chlorite and the mixed layersilicates indicates that sherd 13/73 did not surpass a temperature of 650◦C duringfiring, since these minerals should have decomposed at this temperature [17], whilemica decomposes only between 900 and 950◦C [18]. The X-ray scan at the bottomof Figure 8 represents a sample from sherd 13/73 heated in the laboratory in airto 930◦C for 24 h. All layer silicate peaks disappeared in the low angle region,only the amphibole peak is still observed. The presence of chlorite and mixed layersilicates in the X-ray diffractogram is decisive in distinguishing Mos-type C fromtypes A, B and D.

Mos-type D, represented by sherd Puca Orqo 13/2 in Figures 4 and 6, showsa rather simple Mössbauer spectrum consisting of only one Fe3+ quadrupole dou-blet and magnetically split hematite. Mos-type D contains no or only very smallamounts of Fe2+. The X-ray scan of sample 13/2 exhibits only small amounts ofmica, indicating a maximum firing temperature near the decomposition tempera-ture of mica between 900 and 950◦C.


Figure 8. X-ray diffraction patterns of the sherd Floral 13/73 as found (top) and after heating in airto 930◦C for 24 h. The identified clay minerals are labelled.

The Mössbauer patterns of type E through H are mainly distinguished from eachother by the varying relative intensities of the individual Fe2+ components. The X-ray scans show few distinctive features, except for the absence of amphiboles intype F (Figure 7) represented by sherd 13/28 and a relatively strong mica 001 peakfor type G represented by sherd 13/61, while only traces of mica are present inthe scans of sherds of Mos-type E and F. Therefore the firing temperature of sherd13/61 must have been below about 900◦C. The RT Mössbauer spectrum of Mos-type H contains no magnetically split components, and only a very small part ofthe Fe3+ species appear as magnetically split hematite at 4.2 K, while the mainmagnetic pattern is a broad background.


4.2. CLASSIFICATION OF FINDS ON THE BASIS OF THEIR MÖSSBAUER

SPECTRA

If sherds of certain archaeological styles also have specific Mössbauer spectra,this can be considered as showing that specific raw materials and firing techniqueswere used in their production. Such a result could be considered as showing thatthe vessels of the different styles had different places of origin. Though the at-tribution of individual Mössbauer spectra to the eight Mos-types described abovemay not always be unambiguous, and despite the limited number of specimensstudied, Table II does, indeed, reveal correlations between the Mos-type and thearchaeological style.

Sherds of type Wacheqsa and Puksha have the very similar types A, and B,Mössbauer patterns typical for oxidising firing. On the other hand, the Qotopukyoand Raku sherds, are all of the strongly reduced Mos-types E and G. The Mosnasherds, which form a compact group in NAA (Section 2) are strongly reduced withMos-types F and G except for one sherd which was oxidised and belongs to MOS-type D. The Mössbauer spectra of the Ofrendas sherds belong to four differentMos-types, both oxidised and reduced, but the reduced form dominates. This maybe taken as an indication that Ofrendas ware was produced in different workshopsby different firing techniques, perhaps at different locations, in agreement withthe NAA data (Figure 1) which reveal widely different element compositions forOfrendas type sherds.

The sherd Floral black (13/76) is a singular find, with only one vessel of thistype being found in the Galería de las Ofrendas. The neutron activation analysisdata also show that it is a singular species in the Galería, though it clusters well

Table II. Frequency of different Mos-types for the individual archaeological styles

Mos-type

A B C D E F G H

Wacheqsa 3 3

Puksha 2 2

Puca Orqo 2 1 2 1 2

Ofrendas 2 3 1 11

Floral 2 1

Dragoniano 1 1 1

Mosna 1 1 4

Raku 4 1

Qotopukyo 4

Sum 5 9 6 4 6 2 23 1


Figure 9. RT Mössbauer spectra of 11 sherds of type Foral black (MOS-type H) showing the overallsimilarity within the group. The first spectrum represents a sherd of the only vessel of Floral blackfound in the Galerıa de las Ofrendas. All other spectra are of surface finds from the settlement ofChavın.

together with a number of surface finds from the settlement of Chavín (Figure 1).The Floral black sherd is also the only specimen of Mos-type H found in theGalería, but the same pattern was observed in the 10 surface finds with whichit clusters together in NAA. The similarity of the Mössbauer spectra of all thesesamples is illustrated in Figure 9.

4.3. ADDITIONAL DETAILS OF THE MÖSSBAUER SPECTRA

The sherds from the Galería de las Ofrendas often have a layer structure with redor brown surfaces and a gray core. Such layer structures are caused by an oxidisingstep at the end of the reducing firing. They are often observed in early pottery fromthe Andes, and generally in prehistoric ceramics [19], perhaps because during firingin covered pit kilns the atmosphere is oxygen-deficient, but at the end the cover ofthe kiln may break open accidentally or be opened intentionally. Oxidation thentakes place starting from the surface and progressing into the interior, but when thetemperature of the sherd becomes too low it may stop before the sherd is oxidisedthroughout. The core and the surface then have different Mössbauer spectra. Whilethe classification by and definition of the different Mos-types has been based onspectra from the interior of sherds, the oxidised surfaces also show spectra of Mos-types A through D. As an example, RT spectra of the surface and the core of sherdMosna 13/25 are shown in Figure 10. The surface has Mos-type D, the core Mos-type G. Oxidation after a preceding reduction is reflected by the Fe3+ quadrupolesplittings around 1.0 mm/s instead of the larger values observed after oxidisingfiring only [18, 20]. The fact that the spectra of the oxidised surfaces and thoseof oxidised cores of Mos-types A through D are so similar, suggests that both are


Figure 10. RT Mössbauer spectra of material from the surface (left) and the core (right) of sherdMosna 13/25. The spectrum of the surface is of Mos-type D, while the spectrum from the core is ofMos-type G. Note the different velocity scales.

Figure 11. Comparison of the RT transmission Mössbauer spectra with the conversion electronMössbauer spectra (CEMS) of material from the cores (left) and the surfaces (right) of the sherdsWacheqsa (13/14) and Raku (13/32). Note the different velocity scales.

due to oxidation at the end of a primarily reducing firing cycle, the cooling ratedetermining whether the oxidation zone reached the interior or not.

Figure 11 shows the RT Mössbauer spectra of material from the cores of thesherds Wacheqsa 13/14 and Raku 13/32 together with the respective conversionelectron Mössbauer (CEMS) spectra of the surfaces. Sherd 13/14 has a reddishcore of Mos-type A and a red surface produced by application of a slip accordingto thin section microscopy. In the core only 11% of mainly ill-defined magnetic


Figure 12. Various Mössbauer spectra of the sherd Ofrendas rojo, 13/55. The RT and low-tempera-ture spectra of material from the core of the sherd are plotted in the top row. In the middle row theRT (left) and 4.2 K (right) spectra of material from the surface are shown. In the bottom row the RTund 4.2 K spectra of the same sample are depicted after heating in air to 800◦C for 24 h. Note thedifferent velocity scales.

components are observed. The CEMS spectrum, which represents the surface to adepth of about 200 nm, contains 38% of well crystallised hematite, confirming thatthe red colour is caused by a slip.

Sherd 13/32 is of Mos-type G. The interior is light gray, but on the surface thereis a thin burnished black layer. The Fe2+ content in the core is 30%, that on thesurface as seen by CEMS is only 21%. Apparently here an oxidation also occurred,but only to a minor extent. The black colour of the surface is probably due tocarbon deposition during the reducing phase of the firing [21]. Carbon and graphitedeposition was recently detected by scanning electron microscopy and XRD inblack ware from Sicán found in a ceramics workshop in Huaca Sialupe [22–24].

Several spectra of sherd Ofrendas rojo 13/55 are shown in Figure 12. All sherdsof the Ofrendas rojo type have red surfaces and gray cores. Both the RT and the


4.2 K spectrum of the core are as expected for Mos-type E with 69% Fe2+ andmere traces of a magnetic component at RT. The RT and 4.2 K spectra of materialfrom the surface are of type D, showing the presence of much hematite in thesurface of the sherd. Oxidation of the surface material at 800◦C in air for 24 h (Fig-ure 12) does not change the Mössbauer patterns significantly. The hematite contentincreases only by about 5% in the RT spectrum, and the trace of Fe2+ disappears.The amount of Fe3+ that is still paramagnetic at 4.2 K is also unchanged, whichshows that the original oxidation of the surface layer must have taken place at ratherhigh temperatures of 800◦C or above. Else laboratory oxidation at 800◦C would beexpected to cause larger changes, for instance in the crystallinity of the hematitethat determines the amount of superparamagnetic hematite in the RT spectrum.

5. Conclusions

The applied methods complement each other well and yield an altogether satisfac-tory classification of the finds from the Galería de las Ofrendas based on differentmaterial properties. Even with the relatively small number of sherds, which wereat our disposal for this study, a grouping is feasible, showing that sherds of thedifferent styles also have different materials properties. It becomes obvious thatthe vessels excavated in the Galería de las Ofrendas are not of local production, butwere brought to Chavín from the neighbouring provinces, presumably as offerings.Neutron activation analysis also shows that ceramics with very similar elementcomposition as that of the finds from the Galería de las Ofrendas are also foundin the settlement of Chavín. This becomes most obvious for the Floral black stylepots, of which only one was found in the Galería, but ten more as surface findsin the settlement of Chavín could be studied in this work. This observation does,however, not necessarily conflict with the notion that the ceramics from the Galeríade las Ofrendas were initially brought to Chavín as offerings, be it on the occasionof a single festivity or over a longer period of time.

The presented ideas result from an interplay of the physical methods and thearchaeological typology (Table II). Detailed investigations of the raw materialsexist only for the finds from the pyramid Tello at Chavín [4], which is youngerthan the finds of the Galería de las Ofrendas. The NAA data for raw materials andceramics found at the Pyramid Tello tally well, in agreement with the notion thatthe ceramics from the Pyramid Tello are of local origin. The fact that none of thesherds found in the Galería resemble those from the Pyramid Tello in their elementcomposition supports the idea that all material in the Galería is non-local in origin.

Unfortunately, no hints could be gained so far from the archaeometric studiesas to the sites or workshops at which the vessels excavated in the Galería de lasOfrendas were manufactured. For this, one would have to make a survey of claydeposits and local ceramics in the possible region in origin, which is a difficult andtime-consuming task. Conclusions as to production methods for the vessels foundin the Galería de las Ofrendas are based on the general picture of the processes


taking place during firing under different conditions. The methods used in makingthe ceramics from the Galería de las Ofrendas appear to have been rather similar tothose reconstructed recently for the ceramic production in several other regions inNorth Peru [15, 18, 20, 25]. At present, a study of more sherds from the Galería delas Ofrendas is under way. Its first results seem to confirm the findings of this work,but eventually the larger data set may yield additional and more comprehensiveinformation of the ceramic material from the Galería de las Ofrendas and its origin.

Acknowledgement

This work was funded by the German Research Council. We are very grateful forthis support.

References

1. Lumbreras, G. L., Chavín de Huántar, Excavaciones en la Galería de las Ofrendas, Vol. 51 ofMat. zur Allgemeinen und Vergleichenden Archäologie, Beck-Verlag, Bonn, 1993.

2. Burger, R., Chavín and the Origins of Andean Civilization, Thames Hudson, London, 1992.3. Kauffmann-Doig, F. and Moreno, M. G., 24 planos de Chavín de Huantar, Vol. 22 of Ar-

queológicas, Museo Nacional de Arqueología, Antropología e Historia del Perú, Lima, 1993,pp. 1–75 + 24 planos.

4. Wagner, U., Wagner, F. E., Stockklauser, A., Salazar, R., Riederer, J. and Kauffmann-Doig, F.,Mössbauer Analysis of Recent Finds from Chavín, Hyp. Interact. 29 (1986), 1113–1116.

5. Wagner, U., Alva, W., Kauffmann-Doig, F., Riederer, J., Salazar, R., Tellenbach, M., Ulbert, C.,Vreeland, J. M., Wagner, F. E. and Müller-Karpe, H., Physical Study of Precolumbian Ceramicsfrom the Andes, In: 1. Fachsymp. der Deutsch-Peruanischen Archäologischen GesellschaftMünchen, Sept. 1985, Konrad Theiss Verlag, 1987, pp. 167–185.

6. Gebhard, R., Kauffmann-Doig, F., Lumbreras, G. L., Riederer, J., Wagner, F. E. and Wagner,U., Mössbauer Study of Ceramic Finds from the Galería de las Ofrendas, Chavín de Huantar,Perú, Hyp. Interact. 2(C) (1997), 6–9.

7. Glascock, M. D., Neff, H. and Vaughn, K. J., Instrumental Neutron Activation Analysis andMultivariate Statistics for Pottery Provenance, In: U. Wagner (ed.), Mössbauer Spectroscopy inArchaeology, Hyp. Interact., topical issue, Vol. I, to be published.

8. Riederer, J., Thin Section Microscopy Applied to the Study of Archaeological Ceramics, In:U. Wagner (ed.), Mössbauer Spectroscopy in Archaeology, Hyp. Interact., topical issue, Vol. I,to be published.

9. Wagner, F. E. and Kyek, A., Mössbauer Spectroscopy in Archaeology: Introduction and Ex-perimental Considerations, In: U. Wagner (ed.), Mössbauer Spectroscopy in Archaeology, Hyp.Interact., topical issue, Vol. I, to be published.

10. Wagner, U., Wagner, F. E., Häusler, W. and Shimada, I., The Use of Mössbauer Spectroscopyin Studies of Archaeological Ceramics, In: D. E. Creagh and D. A. Bradley (eds.), Radiation inArt and Archaeometry, Elsevier, Amsterdam, 2000, pp. 417–443.

11. Häusler, W., Unpublished results, 2002.12. Moore, D. M. and Reynolds, Jr., R. C., X-ray Diffraction and the Identification and Analysis of

Clay Minerals, Oxford University Press, 1989, p. 223.13. Dollase, W. A., Mössbauer Spectra and Iron Distribution in the Epidote-Group Minerals, Z.

Kristallographie 138 (1973), 41–63.


14. Fehr, K. T. and Heuss-Assbichler, S., Intracrystalline Equilibria and Immiscibility along theJoin Clinozoisite-Epidote: an Experimental and 57Fe Mössbauer Study, Neues Jahrb. Miner.Abh. 172 (1997), 43–63.

15. Tschauner, H. and Wagner, U., Pottery from a Chimú Workshop Studied by Mössbauer Spec-troscopy, In: U. Wagner (ed.), Mössbauer Spectroscopy in Archaeology, Hyp. Interact. 150,topical issue, Vol. II, 2003.

16. Moore, D. M. and Reynolds, Jr., R. C., Identification of Clay Minerals and Associated Minerals,In: X-ray Diffraction and the Identification and Analysis of Clay Minerals, Oxford UniversityPress, 1989, pp. 202–240.

17. Barnhisel, R. I. and Bertsch, P. M., Chlorites and Hydroxy-Interlayered Vermiculite and Smec-tite, In: J. B. Dixon and S. B. Weed (eds.), Minerals in Soil Environments, 2nd edn, Soil Scienceof America Book Series, Soil Science Society of America, Madison, Wisconsin, USA, 1989,pp. 729–788.

18. Häusler, W., Firing of Clay: Studied by X-ray Diffraction and Mössbauer Spectroscopy, In:U. Wagner (ed.), Mössbauer Spectroscopy in Archaeology, Hyp. Interact., topical issue, Vol. I,to be published.

19. Wagner, U., Gebhard, R., Häusler, W., Hutzelmann, T., Riederer, J., Shimada, I., Sosa, J.and Wagner, F. E., Reducing Firing of an Early Pottery Making Kiln at Batán Grande, Peru:A Mössbauer Study, Hyp. Interact. 122 (1999), 163–170.

20. Shimada, I., Häusler, W., Hutzelmann, T. and Wagner, U., Early Pottery Making in NorthernCoastal Peru. Part I: Mössbauer Study of Clays, In: U. Wagner (ed.), Mössbauer Spectroscopyin Archaeology, Hyp. Interact. 150, topical issue, Vol. II, 2003.

21. Noll, W., Anorganische Pigmente in Vorgeschichte und Antike, Fortschr. Miner. 57 (1979),203–263.

22. Shimada, I. and Wagner, U., Peruvian Black Pottery Production and Metalworking: A MiddleSicán Craft Workshop at Huaca Sialupe, MRS Bulletin 26(1) (2001), 25–30.

23. Stanjek, H. and Häusler, W., Basics of X-ray Diffraction, In: U. Wagner (ed.), MössbauerSpectroscopy in Archaeology, Hyp. Interact., topical issue, Vol. I, to be published.

24. Froh, J., Archaeological Ceramics Studied by Scanning Electron Microscopy, In: U. Wagner(ed.), Mössbauer Spectroscopy in Archaeology, Hyp. Interact., topical issue, Vol. I, to bepublished.

25. Hayashida, F., Häusler, W. and Wagner, U., Technology and Organisation of Inka Pottery Pro-duction in the Leche Valley. Part I: Study of Clays and Unfired Sherds, In: U. Wagner (ed.),Mössbauer Spectroscopy in Archaeology, Hyp. Interact. 150, topical issue, Vol. II, 2003.

Download - Mössbauer Study of Ceramic Finds from the Galería de las Ofrendas, Chavín de Huántar

Top Related