www.nature.

com/scientificreports

OPEN Characterization of Castellani

nineteenth‑century gold jewellery
by in situ micro‑XRF spectroscopy
Rosarosa Manca1,10*, Simona Scrivano2,3, Chiara Manfriani4,11, Francisco J. Ager Vàzquez3,5,
Inés Ortega‑Feliu3,6, Marco Ferretti7, Miguel Angel Respaldiza3,8 & Marco Benvenuti1,9

The material characterization of nineteenth-century artifacts is of great interest, due both to the
breakthrough technological advances and to the unprecedented spread of forgeries of antiquities
which took place in that period. However, this type of artifacts has been largely overlooked in the
past. In this paper we present the compositional analysis of gold jewels by the Castellani, one of the
most important families of goldsmiths in nineteenth-century Europe. The use of a portable micro-XRF
spectrometer, specifically developed for jewellery analysis at the Centro Nacional de Aceleradores
(Seville), allowed us to analyse, in a completely non-invasive way, the alloys, joining techniques and
surface treatments of the jewels of the Castellani collection at the National Etruscan Museum of Villa
Giulia in Rome. The addition to the spectrometer of a second X-rays detector with a zinc filter allowed
us to check the possible presence of low amounts of cadmium, a metal added to gold soldering only
from the nineteenth century and often used in authenticity studies. Moreover, the relative intensities
of the Au X-ray lines were studied in order to check non-invasively the presence of surface enrichments
in gold.

The nineteenth century was characterized by great cultural and technological transformations. Important inno-
vations were taking place in the frame of the Industrial Revolution and new materials and techniques became
available also for artists and craftsmen. At the same time, a deep fascination for ancient civilizations, inspired
by the important and numerous archaeological discoveries of the time, spread all over Europe and the revivalist
(or historicist) style reached a great popularity. In particular, jewellery design witnessed a time of significant
transformation in the second half of the ­century1.
Despite their historical significance, nineteenth-century jewels have been overlooked in archaeometric and
heritage science studies to date and an extensive knowledge of the alloys, joining techniques and surface treat-
ments used by the goldsmiths of this period is currently missing.
One of the reason of this lack of data is to be identified in the difficulties related to the analysis of gold jewels,
which requires: (a) the use of non-invasive techniques; (b) the production of quantitative results, since the alloys
are invariably made of gold, silver and copper as main components; (c) the capability to investigate micro-areas;
(d) the use of portable equipment, since it is often impossible to transport the jewels to a laboratory. Portable
X-Ray Fluorescence spectrometers with a micro beam (micro-XRF) fulfil all these requirements and their recent
development has paved the way for new exciting opportunities in the study of gold j­ewels2–5. In this study, we
used the micro-XRF developed by the Centro Nacional de Aceleradores (CNA) of S­ eville4, which, to the authors’
knowledge, is the only one specifically optimised for the study of ancient gold jewels and currently dedicated
to it (Fig. 1). An ad-hoc modification was done to the equipment to better achieve the aims of this study, which

1
Earth Sciences Department, Università degli Studi di Firenze, Via La Pira 4, 50121 Firenze, Italy. 2Laboratorio
de Rayos X, CITIUS, Universidad de Sevilla, Av. de Reina Mercedes 4b, 41012 Sevilla, Spain. 3Centro Nacional de
Aceleradores, Universidad de Sevilla–CSIC-Junta de Andalucía, c/ T. A. Edison 7, 41092 Sevilla, Spain. 4Dipartimento
di Studi Umanistici, Università degli Studi di Roma Tre, Via Ostiense 234, 00146 Roma, Italy. 5Departamento
de Física Aplicada I, Universidad de Sevilla, c/ Virgen de África 7, 41011 Sevilla, Spain. 6Departamento de Física
Aplicada III, Universidad de Sevilla, Camino de los Descubrimientos s/n, 41092 Sevilla, Spain. 7Istituto di Scienze
del Patrimonio Culturale, Consiglio Nazionale delle Ricerche, Via Salaria km 29.300, 00015 Montelibretti,
Roma, Italy. 8Departamento de Física Atómica, Molecular y Nuclear, Universidad de Sevilla, Av. Reina Mercedes
s/n, Sevilla, Spain. 9Istituto di Geoscienze e Georisorse, Consiglio Nazionale delle Ricerche, via G. La Pira, 4,
50121 Firenze, Italy. 10Istituto di Scienze del Patrimonio Culturale, Consiglio Nazionale delle Ricerche, Via Cardinale
Guglielmo Sanfelice 8, 80134 Napoli, Italy. 11DAGRI, Università degli Studi di Firenze, Via San Bonaventura 13,
50145 Firenze, Italy. *email: [email protected]

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 1

Vol.:(0123456789)
 www.nature.com/scientificreports/

Figure 1.  Top left: Detail of the analysis area of the CNA portable micro-XRF used in this study. The tip of the
two detectors (with and without the zinc filter), the polycapillary lens and a LED ring to illuminate the sample
are shown. Top right: close-up showing the polycapillary on the right, the detector with the zinc filter applied on
a plastic support on the left, and the detector without filters in the background. Bottom: Images of the Castellani
jewels seen under the microscope of the micro-XRF. The red circle (100 µm) indicates the area of analysis (note
that the real beam spot size is 30 µm).

is focused on the production of the Castellani, a family of influential goldsmiths, collectors and antiques deal-
ers from Rome. Making remarkable jewels inspired by archaeological models (especially Etruscan, Greek and
Roman ones, but also medieval and Renaissance ones), they introduced the fashion of ‘archaeological jewellery’
in nineteenth-century E ­ urope6–11.
This study aims at providing a first, broad and consistent set of compositional data of revivalist jewels made
in the nineteenth century. The acquisition of such a data set will expand our understanding on the kind of alloys,
joining techniques, and surface treatments chosen by the Castellani to replicate ancient jewels. Such information,
acquired in a completely non invasive way, would be significant not only for the history of jewellery, but also to
provide additional tools for the detection of the numerous forgeries produced in this fascinating and complex
period, in which the legal and illegal antiques market reached unprecedented p ­ roportions12.

Castellani goldsmithing techniques. The activity of the Castellani - Fortunato Pio (1794–1865) and his
sons Alessandro (1823–1883) and Augusto (1829–1914) - dates between 1814 and 1930, with its peak between
the 1850s and the 1880s. They devoted part of their activity to investigate archaeological jewels, such as those
of the recently discovered Regolini–Galassi tomb and of the collection of the Marquess Campana, and the tech-
niques to reproduce them.
Castellani jewellery has been studied from an art-history perspective from the ­1970s6–9, but technical aspects
related to their activity have been only marginally ­investigated10,11,13,14, so that there are many open questions
about their production practices which can be addressed by compositional analyses.
First of all, it is not known which kind of alloys the Castellani workshops were using and whether they chose
different alloys according to specific techniques or styles. It is interesting in this respect, that the Castellani had
been exempted from the obligation, imposed on all the other Roman goldsmiths, to certify the purity of the gold
alloys they ­used15. Moreover, the Castellani were known for the use of surface treatments, called ‘colouring’16–18.
These treatments consisted in the application of etching solutions to remove copper and silver from the gold
alloy to leave the surface enriched in ­gold19–21. The possible presence of surface enrichments on Castellani jewels
is certainly a relevant aspect to be considered when performing compositional analysis and in this paper we
present the use of a non-invasive methodology, based on the study of the relative intensities of the X-ray lines
of gold in the XRF spectra, to extrapolate information about the possible presence of compositional gradients
from the surface to the subsurface of Castellani jewels.
Another technique deeply investigated and applied by the Castellani is granulation, i.e. the application of
minuscule gold spheres (granules) on a gold foil to create linear decorations or ­patterns22,23 (Fig. 1). The main
difficulty of the technique consists in the ability to solder efficiently the granules to the foil without leaving
residues. The Etruscans were undisputed masters of it, especially thanks to the use of the copper salts solder-
ing ­technique24–27. However the latter had been lost by the nineteenth century, when the Castellanis started to

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 2

Vol:.(1234567890)
www.nature.com/scientificreports/

dedicate their attention to granulation in the attempt of reproducing the ancient masterpieces and it is not known
with certainty which technique did they u ­ se13.
As for the soldering techniques, it is possible that the Castellani were among the firsts to use cadmium-con-
taining ­solders10,28,29, which were indeed introduced in jewellery in the 1860s, then becoming common only in
the twentieth c­ entury20,29,30. The detection of cadmium in a gold jewel (in amounts between c. 2–15 wt%29,31) has
received much attention, because it can be used as a proof of forgery in authenticity ­studies28–30. This is relevant
also because, as discussed by many a­ uthors6,9,15,28,32, it cannot be excluded that the Castellani, just as many other
revivalist artists, had been involved in the production of forgeries. However, the detection of cadmium offers
some technical issues, since the characteristic peaks of cadmium can be hidden by other, interfering peaks in
SEM–EDS, PIXE and XRF ­analysis29. In the framework of this study, a second X-rays detector with a filter was
added to the CNA equipment in order to delete the sum peaks of gold which can hide the main characteristic line
of cadmium. The use of secondary filters, well established in Proton Induced X-ray Emission (PIXE) ­analysis33–35,
has been only recently proposed for XRF spectroscopy by some of the authors of this paper for the detection of Pd
in ancient gold ­jewels3. Their application for the detection of cadmium in gold matrices is presented in this paper.

Materials
The collection of Castellani jewels at the National Etruscan Museum of Villa Giulia in Rome (Villa Giulia here-
after) was identified as the most appropriate for this study, because it is the only one passed directly from the
Castellani workshop to a public museum without being manipulated by others, and because it offers the widest
possible range of jewels produced by t­ hem8. The collection was gathered by Augusto Castellani between the 1870s
and the 1880s, but the date of production of the individual jewels is unknown.
Castellani jewels at Villa Giulia are divided into eight groups, following Augusto’s classification of Italian jewel-
lery ­history36 (Table 1). The selected jewels pertain to all the different periods identified by Augusto, excluding
the ʻPrimigenioʼ, which does not include gold jewels. Forty-three jewels were selected for the analysis following
stylistic and technological criteria, in order to cover the widest possible range of artifacts (Table 1, Fig. 2). Spe-
cifically, the jewels were selected considering the presence of: the Castellani ‘CC’ mark; decorations made with
granulation; precious stones and/or enamels. Moreover, the selected jewels include: exceptional jewels made in
a single copy; commercial jewels reproduced in many examples; jewels copied from a known ancient prototype;
models to be shown to customers; and groups of jewels with similar stylistic features.
The joining areas were analysed in 24 jewels.

Analytical methods
Improvement of the CNA micro‑XRF set‑up. The selected jewels were analysed with the CNA portable
micro-XRF system specifically modified for this study to improve the detection of trace elements in the gold
matrix (Fig. 1). The first version of the CNA micro-XRF is fully described by Scrivano et al.4, while some minor
modifications are reported in Ortega-Feliu et al.3. The main characteristics of the equipment used in this study
are summarised in Table 2.
The analysis of single wires, granules and joining areas is made possible by the use of polycapillary optics
which focus the primary X-ray beam down to 30 μm (lateral resolution measured using a copper ­grid4) and of
a pointing system equipped with a microscope. The jewel under analysis is placed on a horizontal, motorised
stage having 1 μm full-step resolution (Fig. 1).
The equipment was modified in order to check the presence of cadmium: a second detector was added to the
system by making a new slot in the measuring structure at 45° take off angle; the detector was equipped with a
20 µm thick zinc foil positioned in front of the beryllium window (Fig. 1). As illustrated by Ortega-Feliu et al.3,
this filter attenuates the gold L X-ray lines and prevents the formation of the related sum peaks (Fig. 5). This
allows to detect unambiguously low amounts (below 1%) of ruthenium (Kα 19.2 keV), palladium (Kα 21.1 keV)
and cadmium (Kα 23.1 keV), whose Kα lines would otherwise overlap with the gold sum peaks at 19.4 keV
(Lα + Lα), 21.1 keV (Lα + Lβ) and 22.8 keV (Lβ + Lβ), respectively. Thanks to this modification, it was possible
to use simultaneously two detectors: one with the zinc filter and one without filters. This allowed to check in a
reliable way the possible presence of low amounts of cadmium in every analysed spot (with the filtered detec-
tor) and, at the same time, detect zinc, use full count rates for quantification and have unaltered relative peaks
intensities for the study of surface enrichments (with the unfiltered detector).
The measurements at Villa Giulia were carried out at 50 kV and 600 µA, usually with 300 s acquisition time.
At least two or three spots were analysed for each part of the jewels that were studied. The quantitative analysis of
the spectra acquired with the unfiltered detector was performed with the elemental sensitivity method through
the QXAS software (International Atomic Energy Agency)38. Three different standards with compositions similar
to the samples were employed to obtain the sensitivity curve (Table 3). The results were normalized to 100% and
expressed in terms of weight percentage. Table 3 shows that the nominal and measured concentrations of the
standards are in good agreement. The mean relative deviation of the measured values from the nominal values
was calculated for a standard not used to construct the sensitivity curve (G4 in Table 3) as 3% for Au, 7% for Ag
and 22% for Cu. Moreover, the standards were analysed daily during the measuring campaign at Villa Giulia
in order to check the stability of the system: thus we collected hundreds of measurements of all the standards,
which were used to evaluate the precision of the analyses. The standard deviation of the average values is below
0.5 wt% for gold, silver and copper for all the standards.
Please note that hereafter weight percentages will be indicated simply with a % for the sake of brevity.

Study of the in‑depth gold concentration. The ratio between the intensities of characteristic X-ray
lines produced by an element can be a valuable source of information in the study of multi-layered metal arti-

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 3

Vol.:(0123456789)
 www.nature.com/scientificreports/

Museum number Description Period CC mark Granulation Date Previous analysis

85002 Fibula with linear granulation Tyrrhenian Yes Yes Unknown –
Fibula with Etruscan inscrip-
85004 Tyrrhenian Yes Yes After 1878 XRF37
tion
85006 Plaque with granulated ducks Tyrrhenian Yes Yes After 1878 XRF37
85011 Disc brooch Tyrrhenian Yes Yes Unknown –
85014 Disc brooch with rays Tyrrhenian Yes Yes After 1878 –
85016 Fibula with lion Etruscan Yes No Unknown XRF37
85017 "Campana" hinged bracelet Tyrrhenian Yes Yes After 1862 –
85019 ‘Bauletto’ earrings Tyrrhenian Yes Yes Unknown SEM–EDS10
85022 Disc brooch ‘millefiori’ Etruscan Yes No Unknown –
Disc brooch with powder
85025 Tyrrhenian Yes Yes Unknown –
granulation
85029 Scarab necklace Etruscan No No After 1858 –
85030 Scarab bracelet Etruscan yes No Unknown –
85032 Scarab pendant Etruscan No Yes Unknown –
85044 Disc earrings with amphora Tyrrhenian Yes Yes Unknown –
Fragment of necklace with Io’s
85045 Magno-Graeco No No Unknown –
heads
85047 Necklace with Io’s heads Magno-Graeco No No After 1862 –
85050 ‘Cuma’ diadem with enamels Tyrrhenian Yes No After 1862 –
85053 Disc earrings with pendants Magno-Graeco Yes No Unknown –
Necklace with Acheloo
85057 Tyrrhenian Yes Yes After 1858 Technological ­exam28
pendant
85060 Brooch with Helios head Etruscan Yes Yes After 1858 –
85063 Model of necklace Modern No No Unknown –
85066 Strap necklace with amphorae Tyrrhenian Yes No Unknown –
85073 ‘Kul Oba’ brooch Tyrrhenian Yes No After 1830 –
85082 Juno’s head earrings Tyrrhenian No No After 1862 –
85120 Duck-shaped earrings Etruscan No No Unknown –
85139 Hairpin with ram’s head Etruscan Yes No Unknown –
85162 Necklace with peltae Roman No No Unknown –
85175 Festoon necklace Roman No No Unknown –
85203 Belt or hair band Tyrrhenian Yes No Unknown –
85232 Cloisonne buckle Medieval Yes No Unknown –
85237 Model of necklace Medieval No No Unknown –
85279 Bracelet with snakes Modern Yes No Unknown –
85292 Brooch in Renaissance style Renaissance Yes No Unknown –
85297 Brooch with engraved emerald Renaissance Yes No Unknown –
85323 Disc brooch AEI Tyrrhenian Yes Yes Unknown –
Paperknife with Queen
85326 Modern Yes No After 1850s –
Margherita
85337 Brooch IO TRIUMPHE Modern Yes No After 1862 –
85369 Hairpin with Juno’s head Tyrrhenian No No After 1862 –
85370 Hairpin with Juno’s head Tyrrhenian No No After 1862 –
85371 Hairpin with ram’s head Etruscan Yes No Unknown –
85418 Hairpin with lapis lazuli Modern Yes No After 1850s –
85419 Hairpin with Imperial eagle Roman No No Unknown –
85494 Enamelled medal UM Modern No No 1893 –

Table 1.  Jewels of the Castellani Collection at Villa Giulia analysed in this study. The period according to
Augusto’s classification of Italian ­jewellery36 is given for each jewel.

facts. This is because the ratio between the intensities of two lines is constant in homogeneous samples (as long as
the acquisition set-up and data elaboration procedure do not change), while variations are expected if composi-
tional gradients are present from the surface to the subsurface in the thickness range investigated by ­XRF39–49. In
this study we propose the use of the ratios Au L2/L3 and Au L3/M to investigate in a non-invasive way the pos-
sible presence of compositional gradients from the surface to the subsurface of the jewels, which could indicate
the use of ‘colouring’ or other superficial treatments.

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 4

Vol:.(1234567890)
www.nature.com/scientificreports/

Figure 2.  Jewels of the Castellani collection at Villa Giulia analysed in this study (photographs taken by the
authors of this study). The museum number is shown close to each jewel.

X-ray tube Rtw MCBM 50-0, 6B W

Anode W
Max voltage; max current 50 kV; 0.6 mA
Primary filter 1 mm thick, aluminium
X-ray beam focusing system Full lens polycapillary 115mls03
Spot size ≈ 30 μm
Working distance 10 mm
Optical microscope VZM 1000i, 2.5×–10×, Edmund optics
Pointing system
2 Micro laser diode modules, 3 mW–635 nm, Edmund Optics
X-ray detector 1 (without filter) SDD, Amptek, X-123
Area; thickness 6 ­mm2; 500 µm
Be window 25 µm thickness
Energy resolution 153 eV FWHM at 5.89 keV
Secondary filter No
X-ray detector 2 (with Zn filter) Super SDD, Amptek, XR-100SDD
Area; thickness 25 ­mm2; 500 µm
Be window 12.5 µm thickness
Energy resolution 137 eV FWHM at 5.89 keV
Secondary filter Zinc filter (20 µm thick)
Excitation/detection geometry 90°/45°
c. 35 kg (whole equipment)
Weight
c. 5 kg (measuring head)

Table 2.  Characteristics of the portable micro-XRF used in this study.

The count rates of the gold X-rays lines were calculated using the open-source software P ­ YMCA50. It must
be noted that PYMCA, when fitting multiple spectra, provides the counts related to the groups of lines L3, L2,
L1 and M, and not of single lines. Therefore, we used the ratios between these groups of lines and not between
single lines. However, the L3-M5 (corresponding to the Lα line in Siegbahn notation) is by far the main line of
the L3 group, covering more than 68% of the total yield of the group, and the L2-M4 (corresponding to the Lβ)
is the dominant line in the L2 group, representing 80% of the yield (­ see50 and references therein). Therefore, the

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 5

Vol.:(0123456789)
 www.nature.com/scientificreports/

Certified composition Measured composition

Au (wt%) Ag (wt%) Cu (wt%)
Standard Au (wt%) Ag (wt%) Cu (wt%) Average σ Average σ Average σ
G1* 95.99 1.99 2.02 96.0 0.1 1.9 0.1 2.1 0.1
G2* 90.01 7.95 2.04 90.2 0.2 7.9 0.1 1.9 0.1
G3* 75 15 10 76.3 0.4 15.2 0.3 8.5 0.5
G4 58.11 28.7 12.19 59.7 0.2 30.7 0.3 9.5 0.1

Table 3.  List of the standards used in this study, reporting both the certified and measured compositions. *The
standards marked with an asterisk were used to construct the sensitivity curve for the quantification.

behaviour showed by the ratios L3/M and L2/L3 well represents those of ratios L3-M5/M and L2-M4/ L3-M5
(i.e. Lα/M and Lβ/Lα).
The thickness range which can be investigated with this method depends on the attenuation length of the gold
lines, defined as the distance into a material where the X-rays drop to 1/e of their incident ­flux51. The attenuation
length of the L3-M5 line of gold in a matrix of pure gold is 3.0 μm, that of the L2-M4 line is 4.5 μm, and that of the
M line is 0.4 μm51. In a real case the sample matrix is not composed by pure gold, but is normally an alloy of Au,
Ag and Cu with a certain degree of p ­ orosity44, so the attenuation length is slightly higher. These thicknesses are
indeed of the same order of magnitude as those theoretically expected for enriched layers (analyses of tumbaga
alloys and the results of experimental tests on mock-up samples available in the literature show that acid attacks
of gold alloys can lead to the formation of enriched layers ranging approximately from 1 to 10 µm, depending on
the starting alloy and treatments ­used19,21,39–41). In addition to the ratio L2/L3, the L3/M ratio is also considered
in this study: in fact, the depth of analysis for the two L lines is quite similar, while it was interesting to use also
the M line to report differences of a very superficial gold content with respect to the sub-layers underneath.
The ratios calculated from 125 XRF spectra acquired from four homogeneous standards (G1, G2, G3, G4,
see Table 3) were taken as a reference, since they have compositions similar to those of the jewels. The following
average ratios and the related uncertainties (3σ) were obtained: L2/L3 = 0.60 ± 0.03 and L3/M = 110 ± 8 (Fig. 7a).
Note that the ratios from samples enriched in gold are expected to be lower than these values, since the less
energetic lines, which are produced in the more superficial layers, are given in the denominator of the ratios.
For example, given a certain intensity of the M line (2.12 keV), the intensity of the L3-M5 line (9.7 keV) in an
enriched sample will be lower than in a homogeneous one, because the overall composition of the surface + sub-
surface will be lower.

Results and discussion

Composition of the gold alloys. The compositions of foils, wires and granules of the Castellani jewels
analysed by micro-XRF are shown in the ternary diagrams in Fig. 3 and reported in Supplementary Table S1
(please note that the complete results of this study are available in Manca’s PhD ­dissertation52).
The gold content ranges from 75 to 99%, i.e. from 18 to 24 carat gold, although a gold content below 80% was
found only in a few spots. Silver is the main alloying metal (average value 7%; maximum value 23%). Copper is
generally present in lower amounts, ranging from 0% up to 8% (average value 2%).
These data showed that most Castellani jewels are made of high carat alloys of gold and silver, with only a
few exceptions which will be discussed below. This is not in agreement with Fortunato Pio’s essay, in which he
stated that he was normally using alloys of about 75% Au and 25% C ­ u16. The occurrence of such a discrepancy
13
had been already hypothesized by O ­ gden on the basis of visual examinations and the SEM–EDS analysis of two
Castellani jewels. The present study allowed to confirm it with strong analytical evidence. A possible explana-
tion for this can be deduced from the same 1826 e­ ssay16, in which Fortunato Pio also stated that “goldsmiths of
ancient civilized nations” were primarily using pure gold or, alternatively, alloys of gold and silver. Therefore, it is
possible that, when they decided to devote their production to ‘archaeological’ jewellery (from about the second
half of the 1830s), the Castellani abandoned the common 18-carat gold rich in copper to use purer alloys of gold
and silver, following what they thought to be the ancient custom.
Despite Castellani’s convictions, it is today known that various alloys were used also in antiquity and that the
Etruscan, for example, did not commonly use high-carat gold. Cesareo and Von H ­ ase37 analysed fifty Etruscan
jewels and four Castellani ones, highlighting that the latter were clearly richer in gold than the Etruscan ones
(Table 4). However, considering the whole range of compositions detected in the present study and by Oliveira
et al.11, Castellani and Etruscan alloys overlap in many cases (Table 4). This confirms the importance of having
a large database to obtain less partial and possibly misleading results.
Having analysed a high number of jewels, a possible correlation between their stylistics and technological
features and the alloys’ compositions was investigated, but it was not possible to identify any correspondence
between them. It was indeed interesting to note that the use of a specific decoration (granulation, filigree, enamel,
etc.) did not correspond to the use of a specific alloy and that the same kind of alloys were used to reproduce
Etruscan, Greek, medieval or Renaissance models without significant distinctions. The highest gold contents
were generally detected in jewels of the ʻTyrrhenianʼ type, the most sophisticated o ­ nes36, often decorated with
granulation, but not all cases follow this criterion (Fig. 3). Similarly, there was no correspondence between the
application of the CC mark and a specific composition: therefore, XRF analysis allowed to ascertain that this

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 6

Vol:.(1234567890)
www.nature.com/scientificreports/

Figure 3.  Ternary diagrams Ag-Au-Cu with the compositions of foils (a), wires (b) and granules (c) of the
Castellani jewels analysed. The jewels are grouped following Augusto Castellani’s classification in historical
­periods36.

Objects References Au (wt%) Ag (wt%) Cu (wt%)

Castellani jewels 76.0–99.4 0.3–23.1 0.2–7.8
Villa Giulia collection (43 jewels) This work 76.0–99.4 0.3–23.1 0.2–7.8
37
Villa Giulia collection (4 jewels) 92.8–96.8 1.4–6.6 0.6–2
11
Maria Pia of Savoy’s casket (26 jewels) 78.7–93.6 6.2–18.9 0.4–4.4
Etruscan jewels (50 jewels) 37
57.0–94.5 5.3–37.5 0.1–10.5
Regolini-Galassi tomb 61.9–83.0 17.0–35.2 0.3–4.9
Bernardini tomb 61.7–94.5 5.3–36.4 0.1–3.9
Castellani collection of
57.0–71.9 23.2–37.5 1.6–10.5
ancient jewels in Villa Giulia
Museo Nazionale, Tarquinia 63.5–75.6 22.8–33.5 1.2–9.0

Table 4.  Ranges of gold, silver and copper contents detected by XRF in this work and previous studies.

mark was not applied with the aim of certifying the gold purity. It is also noteworthy that the Castellani used
very pure gold, despite the fact that they were exempted to mark its fineness. Even the two models of necklace
85063 and 85237, which are not complete jewels but just samples to be shown to customers, are made with high-
purity gold. Moreover, the fact that the analysed jewels have variable compositions and that they do not cluster
in separate groups, showing a practically continuous distribution in the range from 80 to 99% gold (Fig. 3), may
suggest that the Castellani were not using fixed, standard alloys, as became common precisely over the nineteenth
­century20,31. Possibly, they were still alloying metals in their own workshops-either mixing prepared bullions
or recycling discarded gold artifacts. It must be also considered that the compositional variability detected in
Castellani jewels may be partially due to the use of ‘colouring’ treatments (see chapter 4.2), which can produce
slightly different superficial concentrations.

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 7

Vol.:(0123456789)
 www.nature.com/scientificreports/

Figure 4.  Ternary diagrams Ag-Au-Cu showing the composition of jewels with similar features. The two
hairpins with ram’s head have the same composition, while the ʻIos headsʼ elements have different compositions
in the two necklaces 85045 and 85047.

The comparison between jewels sharing similar stylistic features showed that earrings and hairpins of the same
couple (Fig. 2, Table 1) have similar compositions (Fig. 4, Supplementary Table S1), suggesting a consistency in
Castellani production. However, the identical ʻIos headsʼ elements decorating both necklaces 85045 and 85047
are made with two different alloys in the two jewels and the lotus-shaped elements have different compositions
in hairpin 85418 and paperknife 85326 (Fig. 4).Therefore, it would be incorrect to imagine a completely uniform
and standardised manufacturing process in the Castellani workshops.
Other types of alloys were also identified in Castellani jewels as highlighted in Fig. 3. Specifically, an 18-carat
gold alloy rich in copper of the type described by Fortunato Pio in the 1826 essay was detected only in the chain
of the model of necklace 85063 (75.5 ± 1.2% Au, 4.8 ± 0.4% Ag and 19.8 ± 0.9% Cu) (Fig. 2). Moreover, the five
gold beads of the jewel showed a peculiar composition: first of all, copper (4.6 ± 1.1%) is more abundant than
silver (2.7 ± 0.1%) in the two external beads; secondly, low amounts of cadmium (c. 0.3%) were detected in all the
beads of the jewel. As shown in Fig. 5, the use of the zinc filter proved indeed to be effective for the unambiguous
detection of cadmium. It is worth of notice that the presence of this element in the foils and not in the joints of
the beads is somehow contradictory, since cadmium is a flux. However, it is unlikely that such a low amount of
cadmium could have had a significant influence either on the melting temperature or on the colour of the alloy.
It can be hypothesized that it is merely a not-intentional impurity rather than a deliberate addition.
Furthermore, specific parts of some jewels showed a relatively low gold content. This was the case of the gold
foil underneath the white enamel in brooch 85292, which was visible and accessible to XRF analysis in some
areas and showed a composition of 76.3 ± 0.4% Au, 16.5 ± 0.3% Ag, 7.2 ± 0.6% Cu, i.e. poorer in gold and richer
in copper than the other foils of the brooch itself and of the other Castellani jewels in general. Similarly, the pins
of brooches and fibulae are made with alloys of 18-carat gold, with an average composition of 73.5 ± 0.7% Au,
14.2 ± 1.3% Ag and 12.3 ± 1.5% Cu (Fig. 6). High copper contents increase the tensile strength of gold alloys and
this may be the reason behind the adoption of such a composition for the pins, which needed to be opened and
closed repeatedly. Only in fibula 85004 the pin is made of 95.5% gold, but this jewel does not seem to be made
to be worn but is rather a ‘study copy’ of an Etruscan fibula of the Campana collection (Louvre, n. 1863 Bj 816).
On the contrary, earrings’ pins have variable compositions, similar to that of the earrings to which they pertain.
Another outlying composition was detected in commemorative medal 85494, in which copper (4.1 ± 0.3%) is
the main alloying metal, while silver is practically absent (Fig. 3). This medal is indeed a rather special and late
jewel by Castellani, minted by the Istituto Poligrafico dello Stato (State Stationery and Mint Office) to be donated
to the army officers and clearly not representative of Castellani’s production. Going on with the exceptions, the
gold settings for the rubies of the pendant-brooch in Renaissance style 85297 contain more copper (6.0 ± 0.9%)
than silver (Ag 2.4 ± 0.2%).

Compositional gradients from the surface to the subsurface. The ratios Au(L3/M) and Au(L2/
L3) calculated for the foils, granules and wires of Castellani jewels are reported in Supplementary Table S1 and
plotted in Fig. 7b: most ratios tend to be lower than those of the homogeneous standards, even if there is not a
complete separation from them. This trend indicates that a surface enrichment in gold is present in many (but
not all) Castellani jewels. Note that the difference between these ‘enriched’ jewels and the standards is higher
than that due to the compositional variations (see the distribution of the four different standards in Fig. 7a).
Examples of selected jewels are displayed in Fig. 7c,d. We can note that all the 18-carat areas analysed (i.e.
brooch pins, the chain of jewel 85063 and the foil supporting the enamel in jewel 85292), which reasonably were
not subjected to treatments of surface enrichment since they have relatively low gold and high copper contents,
do plot in the same area of the standards (Fig. 7c). This result provides an indirect validation of the method. The
gold peaks ratios of the jewels overlap with the standards’ area also in the case of: foils with higher copper than
silver content, such as those of commemorative medal 85494; the jewels in which the granulation was glued with

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 8

Vol:.(1234567890)
www.nature.com/scientificreports/

Figure 5.  Top: Spectra from the gold standard G1 (95.98% Au, 1.99% Ag, 2.02% Cu). Note that, in the
spectrum acquired with the zinc filter (red trace), the gold Lα and Lβ peaks are strongly attenuated and the
related sum peaks are removed. Bottom: Spectra from a bead of model of necklace 85063 (95.7% Au, 2.1%
Ag, 2.1% Cu, 0.3% Cd). Low peaks around 21 and 23 keV can be discerned in the spectrum acquired without
the filter (black trace) and it is not possible to determine whether they are sum peaks or not. Using the filter
(red trace) the peak at 21 keV disappears, indicating that it was the Au(Lα + Lβ) sum peak, while the peak at
23.10 keV remains visible, indicating that low amounts of cadmium are present.

an organic resin, i.e. 85057 (with one outlier) and 85323; and granulated plaque 85006 (Fig. 7c). Therefore, we
can deduce that these jewels do not have a surface enriched in gold.
On the contrary, a rather clear separation from the area of the standards can be appreciated for other jewels,
such as the granulated fibula 85004, the strap necklace 85066, the bauletto earrings 85019 and the ‘Kul Oba’
brooch 85073 shown in Fig. 7d. An interesting example is that of brooch 85073, which has a very inhomogene-
ous composition. This could be attributed to the differential action of acidic solutions which acted unevenly on
the surfaces, leaving areas with different degrees of ‘corrosion’ both in terms of composition and thickness. The
fact that the peak area ratios do not overlap with those of the standards is compatible with this interpretation.
In addition to this, it is also interesting to note that a ‘Kul Oba’ brooch in private collection was analysed by
SEM–EDS by ­Ogden13 and showed a gold content above 99%. SEM–EDS investigates smaller thicknesses than
XRF (approximately 5 µm vs 50 µm in gold ­matrices53): therefore, the comparison between XRF and SEM–EDS
results provides interesting information about the presence of compositional gradients in the (sub)superficial
­layers53,54. The same behaviour was observed for the other two Castellani jewels analysed by SEM–EDS in previ-
ous studies: ­Ogden13 and Virgili et al.10 report a composition of c. 99% for an earring with Juno’s head in private
collection (of the same type of Villa Giulia 85082) and the ‘bauletto’ earring 85019, respectively. On the contrary,
XRF analysis made in this study revealed lower gold contents in earrings 85082 and 85019 (between 93 and 98%).
Going back to brooch 85073, it was interesting to note that its pin is made of a non-enriched, 18-carat alloy,
suggesting that the surface treatment was done before the addition of the pin. The latter consideration is valid
also for the other brooches with an 18-carat pin.

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 9

Vol.:(0123456789)
 www.nature.com/scientificreports/

Figure 6.  Compositions of foils (crosses) and pins (triangles) of Castellani brooches.

Figure 7.  Plots of the gold X-rays lines ratios L3/M vs L3/L2. The ratios obtained from the homogeneous
standards are shown in plot (a). The grey rectangle indicates in all plots the area covered by the standards
(average ± 3σ) and used as reference. The ratios obtained from foils, granules and wires of all the Castellani
jewels analysed are shown in graph (b). A selection of spots from graph (b) is given in graphs (c,d), showing
jewels with ratios lower than those of the standards (c) or comparable to them (d).

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 10

Vol:.(1234567890)
www.nature.com/scientificreports/

Jewel Area Au (wt%) Ag (wt%) Cu (wt%)

Foils (average value) 97.6 1.6 0.8
Possible soldering residues where granules are missing 94.5 4.6 0.9
Possible soldering residues where granules are missing 92.5 5.7 1.9
85004
Granules (average value) 98.1 1.2 0.7
Joining area: granule + granule 90.9 7.9 1.1
Joining area: granule + granule 91.3 7.6 1.1
Granules (average value) 96.9 1.6 1.6
Joining area: granule + granule 92.8 5.5 1.6
Joining area: granule + granule 91.6 6.7 1.7
Joining area: granule + granule 91.9 5.9 2.3
Joining area: granule + granule 94.7 3.0 2.3
85006
Joining area: granule + granule 92.0 5.6 2.3
Joining area: granule + granule 89.7 8.5 1.9
Joining area: granule + granule 91.6 6.5 1.9
Joining area: granule + granule 95.8 2.7 1.5
Joining area: granule + granule 94.1 4.3 1.6
Granules in the hair (average value) 96.7 2.7 0.6
Joining area: granule + granule 89.3 9.3 1.4
85060 Joining area: granule + granule 87.6 11.9 0.5
Granules in the eye (average value) 98.5 1.1 0.4
Joining area: granule + granule 80.1 16.2 3.7

Table 5.  Composition of some of the joining areas analysed in Castellani granulation. The average
compositions of granules and/or foils of each jewel are reported for comparison and are in italics.

An ambiguous behaviour of the gold peaks ratios was observed in other jewels, since some spots fall in the
same range of the standards, while others do not. This could be explained considering that surface treatments
normally have inhomogeneous effects; therefore, some spots may have been affected more than others. Simi-
larly, the fact that the two ratios do not always behave in the same way-meaning that in some cases the L3/M
ratio overlap with that of the standards, while the L2/L3 does not, or vice versa-can be explained considering
that each ratio has a different range of sensitivity and that different thicknesses may have been affected by the
‘colouring’ treatment.
Finally, the fact that wires and granules show the same behaviour as foils is noteworthy, since it indicates that,
when detected, the surface enrichment in gold is due to a deliberate treatment applied to the whole jewel and
not to the processes involved in the production of the foils (repeated cycles of hammering and annealing), which
indeed can also result in a slight gold e­ nrichment55,56. At the same time, it cannot be completely excluded that a
simple pickling may be responsible of the compositional gradient identified. However, the weak acids used for
pickling are more efficient in the removal of copper than in that of ­silver57, while the depletion of silver seems
to be the main responsible of the gold enrichment occurring in Castellani jewels. Therefore, the application of
strong etching treatments (‘colouring’) seems more likely.

Soldering techniques. An increase of the silver content and a correspondent decrease in gold were
detected in all the joining areas analysed (see Supplementary Table S1 and examples in Table 5). Copper has
a variable behaviour: often a minor increase was reported together with that of silver, while in some cases the
amounts of copper remain constant. This result indicates the use of brazing alloys rich in s­ ilver27. An increase
of the copper content alone-which indicates the use of copper s­ alts27-was never detected, confirming that the
Castellani did not know this ­technique13.
It is important to stress that the compositions obtained are not the original ones of the soldering alloys used
but are rather the result of the interaction between them and the parts to be soldered. However, a gross residue
of solder with a silvery colour was visible to the naked eye under the pin of bauletto earring 85019B and allowed
us to characterize a composition which is probably similar to the original brazing alloy: 39.8% Au, 45.2% Ag,
15.1% Cu, corresponding to a melting point of c. 850 °C (Fig. 8)21.
No other metals were detected in the solders of Villa Giulia’s jewels. In particular, cadmium was never detected
in the joining areas of the Castellani jewels analysed. Therefore, the possible use of cadmium-containing solders
by the Castellani, hypothesized by ­Ogden13, was not confirmed by the present study. This result has interesting
implications concerning the involvement of Castellani in the production of forgeries. Specifically, the manu-
facture of a forged bauletto earring today at the Antikesammlung of Berlin (inv. GI 196) was attributed both by
Formigli and H ­ eilmeyer28 and O ­ gden13 to the Castellani on the basis of technological features. A content of c.
6% Cd was detected by SEM–EDS in the solders of this forged e­ arring28. Considering the results of the present
research, it seems unlikely that the Castellani decided to use cadmium solders for the production of forgeries
when they were not using them for their legal, revival production, especially considering Castellani’s attention
to the reproduction of ancient techniques. However, it must be remembered that Castellani production is not

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 11

Vol.:(0123456789)
 www.nature.com/scientificreports/

Figure 8.  Ternary diagram showing the compositions and related melting temperatures of the alloys of necklace
85066 and earring 85019B. The red star indicates the composition of the residue of brazing alloy visible under
the pin of ‘bauletto’ earring 85019B and shown in the image. Liquidus areas are taken from S­ cott21.

attributable to a single centre, but was arranged in many w ­ orkshops58. Specifically, the jewels today at Villa
Giulia were part of Augusto’s collection and it is, therefore, likely that the results of the present study are more
representative of Augusto’s activity, while ­Ogden13 attributed the forged earring to the workshop of Alessandro
Castellani and Giacinto Melillo in Naples.
For the moment the only other data available in the literature on Castellani’s solders are those reported by
­Ogden13 for two jewels in private collection analysed by SEM–EDS: an alloy of gold, silver and copper in roughly
equal proportions with 5.5% zinc and 1% cadmium (likely to be an impurity of zinc) was identified in a Kul Oba
brooch, while an alloy with 80% gold, 15% copper and 5% silver was detected in a Juno-head earring. These
solders are clearly different from those of Villa Giulia jewels: this may indicate that the latter jewels have been
made in different workshops, possibly controlled by Alessandro-more active abroad-rather than by Augusto,
but this theory remains largely speculative.
Regarding granulation, it is worth to be stressed that the granules as well are soldered with brazing alloys rich
in silver. Only two exceptions were identified: in the Acheloo pendant 85057 and the ‘AEI’ brooch 85323 the all-
over granulation is not soldered, but accurately glued with a transparent resin (Fig. 9). Formigli and H ­ eilmeyer28
already observed the use of a “hardened organic glue” in pendant 85057 and described it as a rudimentary tech-
nique, never seen on other jewels, and probably to be attributed to one of the first experiments of granulation
by Castellani. There is no mention in the literature of the use of this method in the AEI brooch 85323, which is
also not dated to the authors’ knowledge, but it is possible to hypothesize that this brooch was made in the same
period as the Acheloo pendant. XRF analysis of the glue was performed in areas where the granulation is missing
and showed the presence of metallic elements probably used as driers of the organic resin (mainly lead, but also
potassium, iron, calcium, copper, zinc, and trace amounts of titanium and arsenic)59,60.

Conclusions
The use of a portable micro-XRF equipment allowed the acquisition of the broadest set of compositional data on
nineteenth-century jewels available to date. Both the base alloys and joining techniques used by the Castellani
goldsmiths were characterised. With only a few exceptions, the Castellani jewels at Villa Giulia were made with
high-carat alloys of gold and silver with copper as a minor component, while the joints were made with brazing
solders of gold and silver, with copper as minor component and no other metals. The use of a detector with a
zinc filter allowed us to exclude without ambiguity the use of cadmium-containing solders in the analysed jewels.
The study of the relative intensities of the L3, L2 and M X-ray lines of gold suggests that a superficial enrich-
ment in gold, compatible with the use of colouring treatments, occurs on many, but not all, Castellani jewels.
Therefore, the new method proposed in this study proved to be a powerful tool to obtain information about
the presence of compositional gradients from the surface to the subsurface of a jewel using solely non-invasive,
superficial analysis.
The analyses presented in this paper provided new insights into technical aspects of the processes of creation
and imitation followed by the Castellani goldsmiths and will be useful in future authenticity studies. Further
research on the materials used by other nineteenth-century goldsmiths, whose productions are still largely
unexplored, as well as on the official standards of the time, will be fundamental to better contextualize Castel-
lani’s production practices. The information obtained in this study will hopefully constitute a solid basis for the
planning of future research on the subject.

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 12

Vol:.(1234567890)
www.nature.com/scientificreports/

Figure 9.  Details of the granulation of pendant 85057 showing the transparent resin in which the granules are
embedded. The XRF spectra obtained from the resin are shown at the bottom.

Data availability
The data generated or analysed during this study are included in this published article and its Supplementary
Information files. Other information and/or raw data are available from the corresponding author on reason-
able request.

Received: 7 March 2022; Accepted: 21 April 2022

References
1. Gere, C. & Rudoe, J. Jewellery at the 1862 exhibition. J. Decor. Arts Soc. 38, 83–105 (2014).
2. MacDonald, C. A. Focusing polycapillary optics and their applications. X-ray Opt. Instrum. 2010, 1–17 (2010).
3. Ortega-Feliu, I. et al. Palladium analysis in gold items from Punic jewellery (Cádiz, Spain). Radiat. Phys. Chem. 167, 108239 (2020).
4. Scrivano, S. et al. In-situ non-destructive analysis of Etruscan gold jewels with the micro-XRF transportable spectrometer from
CNA. J. Archaeol. Sci. Rep. 16, 185–193 (2017).
5. Vittiglio, G. et al. A compact μ-XRF spectrometer for (in situ) analyses of cultural heritage and forensic materials. Nucl. Instrum.
Methods Phys. Res. Sect. B Beam Interact. Mater. At. 213, 693–698 (2004).
6. Bordenache Battaglia, G., Gajo, M. G. & Monsagrati, G. Castellani. Dizionario Biografico degli Italiani, Vol. 21 (1978).
7. Munn, G. C. Castellani and Giuliano: Revivalist Jewellers of the 19th Century (Rizzoli, 1984).
8. Moretti Sgubini, A. M. La collezione Augusto Castellani (L’Erma di Bretschneider, 2000).
9. Weber Soros, S. & Walker, S. Castellani and Italian Archaeological Jewelry (Yale University Press, 2004).
10. Virgili, V., Carraro, A. & Ferro, D. Cluster analysis: A precious tool for study antique and Etruscan jewellery from Castellani collec-
tion. In Proc. CMA4CH 2nd International Meeting on Multivariate Analysis and Chemometry for Cultural Heritage and Enviroment,
Torino, 69–70 (Marco Valerio publisher, 2008).
11. Oliveira, M. J. et al. The jewellery from the casket of Maria Pia of Savoy, Queen of Portugal, produced at Castellani’s workshop.
ArchéoSciences 33, 265–270. https://​doi.​org/​10.​4000/​arche​oscie​nces.​2308 (2009).
12. La Niece, S. Forgeries and public collections. ArchéoSciences 33, 329–333. https://​doi.​org/​10.​4000/​arche​oscie​nces.​2419 (2009).
13. Ogden, J. Revivers of the lost art: Alessandro Castellani and the quest for classical precision. In Castellani and Italian Archaeological
Jewelry (ed. Donati, M.) 181–198 (Yale University Press, 2004).
14. Donati, M. The workshop: Models, tools, and newly discovered designs. In Castellani and Italian Archaeological Jewelry (ed. Donati,
M.) 129–151 (Castellani and Italian Archaeological Jewelry, 2004).
15. Caruso, I. Il principio della “falsicabilitá” nella bottega Castellani: copie controllate e libera circolazione. In De re metallica (ed.
Agricola, G.) 129–139 (L’Erma di Bretschneider, 2006).

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 13

Vol.:(0123456789)
 www.nature.com/scientificreports/

16. Castellani, F. P. Ricerche chimico-tecnologiche sul colorimento detto Giallone delle manifatture d’oro, con alcun cenno sulle
dorature dei bronzi; memoria dell’orefice F. P. Castellani letta all’accademia dei Lincei il dì 10 agosto 1826. G. Arcadico Sci. Lett.
Ed. Artif. 32, 62–90 (1826).
17. Castellani, A. A Memoir on the jewellery of the ancients. In Proc. Meetings of the Archaeological Institute, July 5 1861. Archaeol. J.,
365–369 (1861).
18. Wigley, T. B. The Art of the Goldsmith and Jeweller: A Treatise on the Manipulation of Gold in the Various Processes of Goldsmith’s
Work, and the Manufacture of Personal Ornaments, etc. For the Use of Students and Practical Men (Charles Griffin and Company,
1898).
19. Grimwade, M. The surface enrichment of carat gold alloys–depletion gilding. Gold Technol. 26, 16–23 (1999).
20. Ogden, J. Age and Authenticity: The Materials and Techniques of 18th and 19th Century Goldsmiths (National Association of Gold-
smiths, 1999).
21. Scott, D. A. Gold and Platinum Metallurgy of Ancient Colombia and Ecuador Vol. 2 (The Getty Conservation Institute, 2012).
22. Lee Carroll, D. A Classification for granulation in ancient metalwork. Am. J. Archaeol. 78, 33 (1974).
23. Wolters, J. The ancient craft of granulation: A re-assessment of established concepts. Gold Bull. 14, 119–129 (1981).
24. Littledale, H. A. P. A new process of Hard-Soldering and Its possible connection with the methods used by Ancient Greeks and
Etruscans. In The Scientific and Technical Factors of Production of Gold and Silverwork: A course of Lectures held at Goldsmiths’
Hall, 1935–36, 44–63 (Goldsmiths’ Company, 1936).
25. Formigli, E. Tecniche dell’Oreficeria Etrusca e Romana. Originali e falsificazioni (Sansoni editore, 1985).
26. Ferro, D., Formigli, E., Pacini, A. & Tossini, D. L. Saldatura Nell’oreficeria Antica (Edizioni Kappa, 2008).
27. Scrivano, S. et al. Compositional and microstructural study of joining methods in archaeological gold objects: Study of joining
methods in archaeological gold objects. X-ray Spectrom. 46, 123–130 (2017).
28. Formigli, E. & Heilmeyer, W. D. Einige Fälschungen antiken Goldschmucks im 19. Jahrhundert. Archaologischer Anz. 299–332
(1993).
29. Meeks, N. D. & Craddock, P. T. The detection of cadmium in gold/silver alloys and its alleged occurrence in ancient gold solders.
Archaeometry 33, 95–107 (1991).
30. Craddock, P. T. Scientific Investigation of Copies, Fakes and Forgeries (Elsevier, 2009).
31. Vitiello, L. Oreficeria moderna. Tecnica-Pratica (Hoepli Editore, 2000).
32. Guerra, M. F. Etruscan gold jewellery pastiches of the Campana’s collection revealed by scientific analysis. In De re metallica (ed.
Agricola, G.) 103–128 (L’Erma di Bretschneider, 2006).
33. Demortier, G. Analysis of gold jewellery artifacts: Characterization of ancient gold solders by PIXE. Gold Bull. 17, 27–38 (1984).
34. Swann, C. P. & Fleming, S. J. Selective filtering in PIXE spectrometry. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact.
Mater. At. 49, 65–69 (1990).
35. Lemasson, Q., Moignard, B., Pacheco, C., Pichon, L. & Filomena Guerra, M. Fast mapping of gold jewellery from ancient Egypt
with PIXE: Searching for hard-solders and PGE inclusions. Talanta 143, 279–286 (2015).
36. Castellani, A. Della Oreficeria Italiana (1872).
37. Cesareo, R. & Von Hase, F. W. Analisi di ori etruschi del VII sec. AC Con uno strumento portatile che impiega la tecnica della
fluorescenza X eccitata da radioisotopi. In Int. Conf. Appl. Nucl. Methods Field Works Art Roma-Venezia 24–29 Maggio 1973 Accad.
Naz. Dei Lincei, 259–296 (1976).
38. International Atomic Energy Agency. WinQXAS-Quantitative X-ray Analysis System for MS Windows Operating System (Version
1.40) (International Atomic Energy Agency, 2002).
39. Cesareo, R., Rizzutto, M. A., Brunetti, A. & Rao, D. V. Metal location and thickness in a multilayered sheet by measuring Kα/Kβ,
Lα/Lβ and Lα/Lγ X-ray ratios. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. At. 267, 2890–2896 (2009).
40. Cesareo, R. et al. Multilayered artifacts in the pre-Columbian metallurgy from the North of Peru. Appl. Phys. A 113, 889–903
(2013).
41. Cesareo, R. et al. Multilayered samples reconstructed by measuring Kα/Kβ or Lα/Lβ X-ray intensity ratios by EDXRF. Nucl. Instrum.
Methods Phys. Res. Sect. B Beam Interact. Mater. At. 312, 15–22 (2013).
42. Trojek, T. & Wegrzynek, D. X-ray fluorescence Kα/Kβ ratios for a layered specimen: Comparison of measurements and Monte Carlo
calculations with the MCNPX code. Nucl. Instrum. Methods Phys. Res. Sect. Accel. Spectrom. Detect. Assoc. Equip. 619, 311–315
(2010).
43. Ferretti, M., Polese, C. & Roldán García, C. X-ray fluorescence investigation of gilded and enamelled silver: The case study of four
medieval processional crosses from central Italy. Spectrochim. Acta B At. Spectrosc. 83–84, 21–27 (2013).
44. Ager, F. J. et al. Reconsidering the accuracy of X-ray fluorescence and ion beam based methods when used to measure the thickness
of ancient gildings. Spectrochim. Acta B At. Spectrosc. 135, 42–47 (2017).
45. Ortega-Feliu, I. et al. Multi-technique characterization of gold electroplating on silver substrates for cultural heritage applications.
Nucl. Instrum. Methods Phys. Res. B Beam Interact. Mater. At. 406, 318–323 (2017).
46. Giurlani, W., Berretti, E., Innocenti, M. & Lavacchi, A. Coating thickness determination using X-ray fluorescence spectroscopy:
Monte Carlo simulations as an alternative to the use of standards. Coatings 9, 79 (2019).
47. Musílek, L., Trojek, T. & Prokeš, R. Techniques for identifying depth inhomogeneities of elemental distribution in materials. Radiat.
Phys. Chem. 167, 108344 (2020).
48. Pessanha, S. et al. Determination of gold leaf thickness in a Renaissance illumination using a nondestructive approach: Determina-
tion of gold leaf thickness by nondestructive approach. X-ray Spectrom. 43, 79–82 (2014).
49. Brocchieri, J., Scialla, E. & Sabbarese, C. Estimation of Ag coating thickness by different methods using a handheld XRF instru-
ment. Nucl. Instrum. Methods Phys. Res. B Beam Interact. Mater. At. 486, 73–84 (2021).
50. Solé, V. A., Papillon, E., Cotte, M., Walter, Ph. & Susini, J. A multiplatform code for the analysis of energy-dispersive X-ray fluo-
rescence spectra. Spectrochim. Acta B At. Spectrosc. 62, 63–68 (2007).
51. Henke, B. L., Gullikson, E. M. & Davis, J. C. X-ray interactions: Photoabsorption, scattering, transmission, and reflection at E =
50–30,000 eV, Z = 1–92. At. Data Nucl. Data Tables 54, 181–342 (1993).
52. Manca, R. Non-invasive, Scientific Analysis of 19th-century Gold Jewellery and Maiolica. A Contribution to technical art History and
Authenticity Studies (Università degli Studi di Firenze, 2021).
53. Blakelock, E. S. Never judge a gold object by its surface analysis: A study of surface phenomena in a selection of gold objects from
the staffordshire hoard: Surface phenomena in gold objects from the Staffordshire Hoard. Archaeometry 58, 912–929 (2016).
54. García-Vuelta, Ó., Montero Ruiz, I. & Villa Valdés, Á. Orfebrería castreña en el Museo Arqueológico de Asturias (Oviedo): Aproxi-
mación a su caracterización arqueométrica y problemas de estudio. Trab. Prehist. 77, 163 (2020).
55. Lechtman, H. Pre-Columbian surface metallurgy. Sci. Am. 250, 56–63 (1984).
56. Sáenz-Samper, J. & Martinón-Torres, M. Depletion gilding, innovation and life-histories: The changing colours of Nahuange
metalwork. Antiquity 91, 1253–1267 (2017).
57. Blakelock, E. Secrets of the Anglo-Saxon goldsmiths: Analysis of gold objects from the Staffordshire Hoard. J. Archaeol. Sci. 72, 44
(2016).
58. Walker, S. Founders, family members, and the firm. In Castellani and Italian Archaeological Jewelry (ed. Walker, S.) 20–65 (Yale
University Press, 2004).
59. Tumosa, C. S. & Mecklenburg, M. F. The influence of lead ions on the drying of oils. Stud. Conserv. 50, 39–47 (2005).

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 14

Vol:.(1234567890)
www.nature.com/scientificreports/

60. Soucek, M. D., Khattab, T. & Wu, J. Review of autoxidation and driers. Prog. Org. Coat. 73, 435–454 (2012).

Acknowledgements
The Museo Nazionale Etrusco di Villa Giulia, and specifically the Director Dr Valentino Nizzo, the curator Dr
Maria Paola Guidobaldi and the conservators Daniela Matticoli and Miriam Lamonaca, are kindly acknowl-
edged for having authorized and facilitated the study of the Castellani Collection in the frame of the Conven-
zione Quadro between the museum and the Earth Sciences Department of the University of Florence (Prot no.
43736, 14/03/2018). We are grateful to Alessandro Pacini, Chiara Pacini and Manuela Petti (Aliseda goldsmith’s
laboratory), to Dr Daniela Ferro and to Prof Marcos Martinón-Torres (University of Cambridge) for their pre-
cious suggestions. RM’s PhD was funded with a “Regione Toscana Pegaso PhD Scholarship 2017”. IOF wishes
to acknowledge the “V Plan Propio de Investigación“ of the University of Seville.

Author contributions
R.M. designed the research project and was in charge of all its phases under the supervision of M.B.; R.M. and
S.S. organized the analytical campaign and defined its aims; S.S., R.M., I.O.F., F.J.A.V. and M.A.R. designed and
made the modifications to the micro-XRF equipment and set it up at Villa Giulia; R.M. and C.M. performed the
analysis and the photographic documentation of the jewels with the help of S.S., F.J.A.V. and I.O.F; R.M. and S.S.
interpreted the data with the contribution of I.O.F. and F.J.A.V.; M.F., R.M. and S.S. performed the study of the
gold lines ratios; R.M. wrote the manuscript and all authors reviewed it.

Competing interests
The authors declare no competing interests.

Additional information
Supplementary Information The online version contains supplementary material available at https://​doi.​org/​
10.​1038/​s41598-​022-​11815-3.
Correspondence and requests for materials should be addressed to R.M.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.

© The Author(s) 2022

Scientific Reports | (2022) 12:7813 | https://doi.org/10.1038/s41598-022-11815-3 15

Vol.:(0123456789)