In this section, the results will be presented and discussed, they are not organized according to the used analytical techniques, but according to the type of information they can provide: the morphological / microstructural study of the artefact, the composition of the alloy, the “reconstruction” of the manufacturing method used to produce the statuette.
3.1 General morphological description
The WB-NT reconstruction confirms that both the figurine and the halo are solid cast. The pedestal is revealed to be a complex hollow structure (Fig. 2). The housing for the halo pins on the side of the square base allows for a mechanical connection: the use of a small amount of hard solder material (Giumlia-Mair, 2012; Azéma, et al., 2017) for this joint cannot be completely ruled out, as there are some points of interconnection between the pins and the wall of the brackets. The contact points in some cases can be identified as earthy concretions showing a lower grey value than the metal, while in specific points the presence of material sharing the same grey tone as the metal is evident. Mineralization phases are certainly present in the interstices left in the pin housing (see Supplementary material S2).
Despite some reconstruction artefacts, such as shadowing effects, the alloy appears homogeneous for the pedestal, the Shiva figurine, and the halo; in fact, the three sections of the statue have relatively similar composition as reported by the ToF-ND compositional analysis (see paragraph 3.2 Alloy Characterization) Fig. 9 and Table 2). In the Shiva figurine, widespread ubiquitous porosity can be observed (Fig. 3): the pores have a rather varied size distribution ranging in size of several hundred µm (500–1000 µm Ø). Considering the spatial resolution limit for cold NT of 480 µm, it seems that neither the halo nor the pedestal is affected by strong porosity.
The statuette is in a good state of conservation considering that it is an archaeological artefact, both from a structural point of view and from the corrosion of the alloy. In the whole artefact, there are no significant fractures except for a single 16 mm long crack inside the belly of the statue, which is likely due to shrinkage of the metal during cooling and does not seem to affect the structural stability of the object.
Hydrogen is present in several mineralization phases and strongly interacts with the neutron, resulting particularly bright in the tomographic images. This allows us to obtain both surface and bulk mapping of the corroded areas. As can be seen in Fig. 3, the presence of mineralized phases is limited to the surface and, in particular, associated with residues of concretions trapped in the undercuts, or in the areas of the figurine richer in finely decorated details.
The same situation can be observed for the pedestal, where the most attenuating phases are identified in the junction between the halo pins and the brackets (for more detailed images, please refer to Supplementary material S2). ToF-ND phase analysis allowed us to identify and quantify some of the main copper alloy alteration products (see Fig. 4 for the inspection locations of ToF-ND analysis). Cuprite is present in low concentrations exceeding 1 wt% only in two analysed measurement areas; nantokite, which can be an indication of active corrosion, is present in only 4 areas of the 21 investigated ranging from 0.1–0.9 wt%. Other phases suggesting active corrosion such as paratacamite and atacamite were not found. Consequently, it seems that the bulk of the statue is not affected by active corrosion.
Chalcocite has not been identified: the undetectability of this phase does not exclude its presence in quantities below the detection limit (0.1 wt%) for the ToF-ND analysis.
3.2 The pedestal
As described in the Introduction, the pedestal is a hollow form consisting of a square base surmounted by a semi-dome structure representing the lotus flower, from which Shiva emerges. There are many examples of statuettes of the similar typology, and many are described in the literature from an archaeological/artistic point of view, but technical studies are rare, especially of the pedestals (some known examples are described in (Craddock & Hook, 2007) and (Dehejia, 2021). Often the pedestal is described as a single piece, either mechanically mounted or cast as one with the figurine of the deity. The tomographic reconstruction highlighted some discontinuities at the height of the upper level of the square base from which "the lotus flower" originates (Fig. 5); these are thickenings of a few millimetres that follow the circumference of the base of the lotus flower. However, this observation is not sufficient to hypothesize that the two parts were cast separately and then assembled, indeed the discontinuities could be attributed to a joint created during the preparation of the wax model of the statuette. The protrusions of wax inside the pedestal, if not removed, could have been faithfully transferred to the mould and finally to the cast bronze (wax on wax joint).
However, as seen in figures (5–7), the square base has four inward extensions directly underneath the plane supporting the lotus pedestal. The sagittal and normal sections show that these supports do not always connect perfectly with the semi-dome structure of the lotus flower but appear to directly support the top plate of the square base (Fig. 8) (see also the pedestal assembly hypothesis in Supplementary material S3, Fig. S6-S8). This could support the hypothesis that the square base was cast independently from the lotus flower which could instead have been cast as a whole with the underlying plate acting as a lid for the underlying base. The two parts could have been joined by welding, and the lid with the lotus flower could have rested perfectly on the four supports. In support of this hypothesis, there are at least four examples of statuettes of the Chola period, in the collection of the Thanjavur Museum which do not have a square base but have the lotus flower connected directly to a rectangular plate (Dehejia, 2021). Other examples of a lotus flower cast separately from a square base can be found as well (Slaczka, et al., 2019; Dehejia, 2021).
WB-NT also allowed us to study the connection system between the Shiva figurine and the pedestal. The feet of the Shiva figurine end in a round flat structure which serves as a base. The statue and the circular footplate were cast together; indeed, no solder lines are visible here. The round footplate has a finer and more compact crystalline grain, probably due to the intense cold mechanical work undergone to fit the housing into the base.
To confirm that this footplate was cast together with Shiva, the microstructure at the interface between the foot and the round plate was observed. In Fig. 6(A) and 6(B) two details of the normal section of the left foot are shown: the first image is a tomographic projection, the second a 3D reconstruction. We observe the presence of two phases: the darker one can be attributed to dendrites with higher tin content. These dendritic structures are more than 500 µm in length on average and up to 2 mm across the interface extending from the foot to the base.
The solidification process of the molten alloy begins in contact with the colder walls of the mould, where a thin layer of crystals is formed, while the dendritic branches, richer in copper (the first component to solidify), propagate inwards. Subsequently, the Sn-rich branches will grow as last elements, on the dendritic structures of the first solidified volume (Scott, 1991). The presence of dendrites across the interface supports the hypothesis that the footplate and feet were cast together rather than soldered. Moreover, the size of these microstructural features suggests that the cooling of the melt was slow enough to allow the formation of elongated dendritic like crystals hundreds of micrometres in size.
The footplate, as seen in the axial projections in Fig. 6 or in detail in Fig. 7 (see also Fig. S9 in Supplementary material S3), rest on the four supports within the lotus flower. The assembly appears to have occurred mechanically by placing the figurine on supports within the flower and then by cold hammering to allow part of the edge to be folded back stabilizing the figurine to the pedestal (Craddock & Hook, 2007). This method is a common practice in modern-day South Indian workshops.
In Fig. 8, we show the 3D segmentation of the three sections of the statuette and a possible representation of the assembly process.
3.2 Alloy characterization
In order to characterize the different portions of the artefact, looking for confirmation of what the WB-NT data suggest, neutron diffraction measurements were performed at multiple locations within the statuette, each location with a gauge volume of 5 mm x 2 mm x 2 mm.
Starting from an accurate determination of the lattice parameter of the Face Centered Cubic (FCC) copper alloy crystalline structure (alpha phase) it was possible to obtain the quantitative analysis of the alloy. In fact, bronze is a substitutional alloy with tin replacing some of the copper atoms, and the lattice parameter of the alloy increases as a function of tin amount with respect to the pure copper value (3.6147 Å). The tin concentration of each sample was estimated by calculating the equivalent tin content as a function of lattice parameter. The equivalent binary Cu-Sn alloy composition was determined using the calibration curves published by Grazzi et al, 2010 (Grazzi, et al., 2010).
The results of the ToF-ND analyses are summarized in Fig. 9 and Table 2. The bronze alloy of the entire artefact is characterized by a low tin content between 2.4–4.6 wt %, and by a lead concentration that varies widely from 2.9 wt% to 12% wt.
Since lead isn't a copper alloying agent (miscibility of 0.1 wt %) and because it is the last element to solidify in a Cu-Sn and Pb system, it is forced to occupy inter-grain positions left by the solidification of the Cu-Sn alloy. Therefore, lead is present in isolated spheroidal accumulations inhomogeneously distributed within the cast metal. Furthermore, its inhomogeneity may also depend on the preparation of the alloy in the crucible (Cantini, et al., 2023; Oudbashi, et al., 2020; Hughes, et al., 1982). For this reason, the various portions of the statue were compared, mainly considering the differences in tin concentration, to later evaluate the differences in lead content.
The halo, certainly cast separately, has an alloy with an average equivalent tin content of 4.6 wt%. The base and the Shiva figurine have very similar alloys (Tables 2 and 3). It is interesting to note that the base has an average Pb content of 3.3 wt%, lower than the other sections of the statuette. This could be attributed to a conscious choice to use a more fluid alloy with a lower melting point to carry out the solid cast: with these characteristics the alloy could have finely filled the details of the moulds of the most valuable portions of the artefact: Shiva figurine and its flaming halo.
Table 2
ToF-ND. Concentration (% wt) of the main elements present in the alloy as derived from Rietveld refinement. The column on the right records the weight percentage content of the equivalent tin in monophasic binary alloy with copper, for each sample.
|
Measurement area
|
SAMPLE
|
Cu (wt%)
|
Er.
|
Sn (wt%)
|
Er.
|
Pb (wt%)
|
Er.
|
Sn (wt%) Binary
|
Er.
|
|
PEDESTAL
|
1
|
93.8
|
0.1
|
2.6
|
0.1
|
3.6
|
0.5
|
2.7
|
0.1
|
|
2
|
94.1
|
0.1
|
2.9
|
0.1
|
3
|
0.9
|
3.0
|
0.1
|
|
3
|
93.5
|
0.1
|
2.8
|
0.1
|
3.7
|
0.5
|
2.9
|
0.1
|
|
4
|
94.6
|
0.1
|
2.5
|
0.1
|
2.9
|
0.4
|
2.6
|
0.1
|
|
RIGHT LEG
|
5
|
88.2
|
0.1
|
3
|
0.1
|
8.8
|
1.4
|
3.3
|
0.1
|
|
6
|
92.2
|
0.1
|
2.5
|
0.1
|
5.3
|
1.1
|
2.6
|
0.1
|
|
7
|
86.1
|
0.1
|
2.7
|
0.1
|
11.2
|
1.6
|
3.0
|
0.1
|
|
HEAD
|
8
|
94.1
|
0.2
|
3
|
0.2
|
2.9
|
1.3
|
3.1
|
0.2
|
|
9
|
90.6
|
0.1
|
3.1
|
0.1
|
6.3
|
0.7
|
3.3
|
0.1
|
|
10
|
90.6
|
0.1
|
3.4
|
0.1
|
6
|
0.6
|
3.6
|
0.1
|
|
11
|
89.8
|
0.1
|
3
|
0.1
|
7.2
|
1.2
|
3.2
|
0.1
|
|
12
|
89.7
|
0.1
|
3.2
|
0.1
|
7.1
|
0.7
|
3.4
|
0.1
|
|
HALO
|
13
|
88.1
|
0.1
|
4.2
|
0.1
|
7.7
|
0.5
|
4.6
|
0.1
|
|
14
|
87.7
|
0.1
|
4.2
|
0.1
|
8.1
|
0.5
|
4.6
|
0.1
|
|
15
|
87.5
|
0.1
|
4.1
|
0.1
|
8.4
|
0.4
|
4.5
|
0.1
|
|
16
|
83.8
|
0.1
|
4.2
|
0.1
|
12
|
0.9
|
4.8
|
0.1
|
|
HEAD (flower, nape)
|
17
|
90.6
|
0.1
|
3.2
|
0.1
|
6.2
|
1.1
|
3.4
|
0.1
|
|
18
|
88.1
|
0.1
|
2.9
|
0.1
|
9
|
1
|
3.2
|
0.1
|
|
RIGHT ARM
|
19
|
88.7
|
0.1
|
2.8
|
0.1
|
8.5
|
0.9
|
3.1
|
0.1
|
|
20
|
93.5
|
0.1
|
3
|
0.1
|
3.5
|
0.9
|
3.1
|
0.1
|
|
21
|
91.8
|
0.1
|
2.9
|
0.1
|
5.3
|
0.5
|
3.1
|
0.1
|
Table 3
ToF-ND Alloy composition results. Average content (wt%)
|
Average concentration (wt%)
|
|
Statuette section
|
Sn (wt%)
|
St. dev
|
Pb (wt%)
|
St. dev
|
Sn (wt%) Binary
|
St.dev
|
|
SHIVA FIGURINE
|
3.0
|
0.2
|
6.7
|
2.3
|
3.2
|
0.2
|
|
PEDESTAL (Square)
|
2.7
|
0.2
|
3.3
|
0.4
|
2.8
|
0.2
|
|
PEDESTAL (Lotus)
|
2.7
|
0.2
|
3.3
|
0.6
|
2.8
|
0.2
|
|
HALO
|
4.2
|
0.1
|
9.1
|
2.0
|
4.6
|
0.1
|
3.3 The pendant
The figurine of Shiva wears a necklace with a small pendant which, according to visual inspection, could be made of silver (now tarnished) possibly alloyed with gold (Fig. 10). From the tomographic images, this decorative element stands out with a very high attenuation coefficient (the grey value is 3 times the value of the body) confirming such hypothesis since both silver and gold have a high neutron absorption cross-section (Ag σscatt = 5 barn; σabs = 63 barn; Au σscatt = 8 barn; σabs = 99 barn).
This assumption is supported by the neutron diffraction analysis, which confirms that its main constituent is Ag or Au (lattice parameter compatible with Ag-Cu or Au-Cu alloy. However, the diffraction method alone, based on lattice parameter determination as discriminating factor was not sufficient to univocally discern the composition. By performing NAA, the elemental composition of the pendant has been determined, yielding the Au and Ag concentration. The composition of the pendant has been determined, taking into account the inhomogeneity of the neutron beam and a correction for neutron self-shielding and gamma self-attenuation (Greenberg, et al., 2011) by simplifying the shape shown in Fig. 10C as a “box” of the same volume. The result shows that the pendant is made of a mixture of Au and Ag with a mass ratio of 1/46 (from a manufacturing practical point of view it can be considered silver).
The ToF-ND phase analysis of the pendant sample volume (point n°8) is characterized by the presence of bronze phase (alpha 1) at 36.3 wt% and by a second major phase at 63.2 wt%. The presence of phase alpha 1 of the bronze is explained as the gauge volume used for the measurements was 5mm x 2mm x 2mm and covered a volume that includes both the pendant and the surrounding bronze. In fact, phase alpha 1 has the same composition as the other points on the body of the Shiva figurine. The other phase is instead representative of the pendant and potentially it is a mixture of Ag-Cu-Au. It is not possible to reach such a lattice parameter value considering an Ag-Au mixture, since the lowest lattice parameter of this alloy is 10% higher than the one obtained. It is then necessary to include copper in the alloy and, because the Ag/Au ratio derived from NAA is 46:1, the Au contribution can be considered negligible. Considering, then, the Ag-Cu alloy (Cu cannot be measured within the pendant using NAA data), it is possible to obtain the measured lattice parameter only by performing quenching within a mould (due to its small size the casting of the pendant can be considered a spontaneous quenching) since, at equilibrium, the maximum copper value hosted in silver is much lower (max 3 at%) providing a higher lattice parameter value. The resulting composition as derived from Pearson (Pearson, 1958) and references within is 95.02 wt% ±0.02 and Cu 4.98 wt% ±0.02 (confirming again as silver from a practical point of view).
3.6 Casting techniques
Tomography data analysis also allows some details about the manufacturing method of the artefact to be inferred.
WB-NT undoubtedly shows that the statue was made in three distinct castings, using the direct lost wax technique: the Shiva statuette was cast together with the round footplate under his feet, while the base and the halo were cast independently. We tried to verify the sprue positions suggested by the literature (Craddock, 2015). According to the traditional South Indian statuette casting technique, the gate system is built in the wax model, at the back of the statue. The main sprues can be located between the shoulder blades or directly in the nape. The mould is then typically buried in a casting pit (Craddock, 2015; Levy, et al., 2007). As far as the mould position of the Shiva figurine is concerned, based on distribution of the pores, it is probable that the legs were oriented upwards, with the round base lying underfoot as a feeder. This is compatible with the reconstructions proposed in the literature (Craddock, 2015), showing a slightly tilted figurine, face down.
The analysis of the set of monochromatic radiographies in the wavelength range between λ = 4.05 Å and λ = 4.20 Å, crossing the (111) Bragg reflection of the Cu-Sn face centered cubic alpha phase shows the presence of anisotropic oligocrystals (mm /sub mm scale) which are constituted by single crystal grains oriented in different spatial directions. This phenomenon is present in the whole artefact: the Shiva figurine, the pedestal, and the halo (Fig. 11a-e) (see also Supplementary material S5 for full size images). They appear as dark spots and the phenomenon is related to a strong coherent scattering effect removing neutrons from the primary beam. Oligo-crystals can certainly be attributed to the bronze microstructure since they are no longer visible in radiographs taken at wavelengths over the (111) family of copper lattice planes (FCC), where the scattering effect of such a phase contributes no more to the attenuation (Su, et al., 2021).
The uneven distribution of single crystal spots shown by ES -NR results for the Halo (Fig. 12), together with neutron diffraction results (lead distribution (Fig S.21 in Supplementary material S6) and related documentation of current production of bronze statuary in Tamil Nadu, allows us to formulate a hypothesis on the position of the Halo mould during casting.
The presence of such oligo-crystals, and dendrites suggests that the entire statuette has undergone a carefully controlled slow cooling. This is in agreement with the custom of placing the mould underground as is still done in the casting tradition of Tamil Nadu (Craddock & Hook, 2007; Levy, et al., 2007).
Figure 13 (A) shows that the presence of large size crystallites interests the entire development of the halo, but their size abruptly decreases at the height of the left forearm. Moreover, it seems that the decrease of the crystals size appears to follow a certain angle. The size of the grains is inversely related to the speed of the solidification process; small crystals implies that part of the halo must have cooled down faster. Based on the models described in the literature, and on the traditional techniques used in contemporary South Indian workshops (Fig. 13 (B)), a casting model for the Halo is proposed (Fig. 13 (C)): the mould was placed in the ground, tilted both towards the front (with the top of the arch pointing down), and sideways, with the left-side sprue only partially buried, thus exposing only that area of the mould to more rapid cooling.