This document describes a study using chemometric analysis to characterize different terracotta finds based on their historical and geographical origins. 20 terracotta samples from 4 archaeological sites were analyzed using ICP-AES, thermogravimetric analysis, and thermomechanical analysis. The results were compiled into a data matrix and principal component analysis was used to classify the samples and correlate their properties with age. The analysis found good correlation between the samples' properties and their known approximate ages, allowing characterization based on multivariate data from multiple analytical techniques.

1. Available online at www.sciencedirect.com
Microchemical Journal 88 (2008) 113 – 120
www.elsevier.com/locate/microc
A chemometric approach to the historical and geographical characterisation
of different terracotta finds
F. Bellanti ⁎, M. Tomassetti, G. Visco, L. Campanella
Rome University, La Sapienza, Pl. Aldo Moro 5, Rome, Italy
Received 24 November 2007; accepted 26 November 2007
Available online 5 December 2007
Abstract
The development of modern analytical instrumental techniques has allowed to solve a lot of problems concerning all kinds of disciplines, but
often we can obtain for the analysed samples very numerous data, obtained by means of analytical techniques. So the effort of the analysts is more
and more paid on the elaboration of a so high number of data. We can find a typical example in the classification and study of archaeological finds.
In the archaeometry, that is the application of scientific methods and analysis techniques to archaeological issues, one of the most important step is
the statistical elaboration of multivariate data obtained by physical and chemical analysis of ancients artefacts. This study was carried out on 20
different pottery samples coming from different periods. The 20 analysed terracotta finds come from 4 different archaeological sites, three Italian
and one Libyan. For the analysis we used ICP–AES, thermogravimetric (TG) and thermomechanical (TMA) techniques; the main technique used
to elaborate the data was the Principal Component Analysis (PCA). We already knew, approximately, the “burning age” of the finds and we had to
check if some eigenvector of Principal Components Analysis could fit with that age. Few milligrams of finds were used for ICP–AES, TGA and
TMA analysis obtaining, after a reduction, a matrix of 11 variables. The results show a good correlation between age and PC1.
© 2007 Elsevier B.V. All rights reserved.
Keywords: ICP; Pottery; Thermoanalysis; Chemometry
1. Introduction Most of archaeometric literature is devoted to the study of
ancient potteries [1], characterized by chemical variables
One of the most interesting application of chemometry is the (oxides, trace elements, e.g.). The most important purpose of
possibility to classify archaeological finds. For this kind of these studies is the determination of their historical and
samples, in fact, frequently the available results have a very geographic origin [2], to obtain information about the used
high number of chemical data from modern analytical materials, the manufacturing techniques, and the cultural and
instrumental techniques as for instance the Inductively Coupled commercial exchanges [3].
Plasma–Atomic Emission Spectrometry (ICP–AES) technique, In a large number of works reported in literature [6–8] the
so we need a tool that can help us to easily obtain from a data classification was conducted using several data provided by the
matrix the information able to classify the samples. ICP–AES spectroscopy; however it is frequently noted that
In this work we show the results of the application of a sometimes ICP–AES data are not sufficient to correctly classify
multivariate analysis to the historical and geographical different pottery samples.
classifications of different terracotta finds, by means of the In this work too we tried to carry out a correct chemometric
elaboration of several instrumental chemical data obtained by classification of potteries [9] originating from different sites (20
different analytical techniques. terracotta finds from 4 different archaeological sites, three
Italian and one Libyan), first of all using only ICP–AES data.
As these data alone were evaluated not suitable for a correct
⁎ Corresponding author. classification of the samples, as shown in this work, we
E-mail address: francesco.bellanti@poste.it (F. Bellanti). introduced in the data matrix used for the classification some
0026-265X/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.microc.2007.11.019

2. 114 F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120
data from thermogravimetric (TG) and thermomechanical powder). To preserve the object an “expert sampling” method
(TMA) technique [10]. This allowed to provide a good was used, looking for bulk material but in low historical
classification. interest position.
All the samples, in the form of non homogeneous fragments,
2. Materials and methods were first carefully ground into homogeneous powder [11].
The experiments were carried out at a heating rate of 10 °C
The potteries tested in the present research come from four min− 1 and under an air flow rate of 100 cm3 min− 1; the
different archaeological sites: terracotta materials were subjected to the thermogravimetric
analysis performed using a Du Pont 951 thermogravimetric
- Archaeological dig on the Libyan Sahara Tadrart Acacus analyzer coupled to a Du Pont thermal analyst 2000 system
massif known as the “Uan Telocat” shelter, (about 5000 under the same atmosphere stream and the same heating rate
B.C.): the finds are five potsherd, with impressed decoration conditions above reported.
obtained using double-pointed comb-like instrument [4]; The ICP–AES analysis was performed by a Jobin-Yvon JY
- Civitella di Chieti dump: three finds are classified as 70 Type III Inductively Coupled Plasma Emission Spectro-
terracotta fragments from fictile statues belonging to the photometer (Horiba Jobin Yvon SAS, Longjumeau, France).
pediment conventionally defined as “type A” and three from The solution of each sample to be analysed was obtained by
pediment defined as “type B”; two other finds belong to the mixing 150 mg of the ground specimen with 1.0 g of lithium
pediment statues at the same dump, although no attributed tetraborate in a graphite crucible and by heating in an oven to
with certainty. All these finds are dating to 1st–2nd centuries 1000 °C for 40 min after slow rising up to 700 °C. The obtained
B.C. [5]; pearl was cooled and than dissolved in 250 ml of aqueous
- Ariccia: three finds originating from different portions of a solution containing 4 ml of HNO3 (65% w/w) and 4 ml of HCl
slightly less than life-size votive statue representing a (37% w/w %), stirring for 5 h. [12]. All reagents were ultra pure
female figure seated on a throne and dated as 3th–2nd from Merck Gmbh.
century B.C.; The thermogravimetric analyses of the Ariccia fictile finds
- Rome: four renaissance potsherds from the excavation of the were performed using a Mettler TG 50 thermobalance
Rome chancery dating to the 15th–16th centuries. connected to a TC 10 A microprocessor and a Swiss dot-
matrix printer (Mettler-Toledo GmbH, Greifensee, Swiss).
All the samples were named following the information in The thermomechanical tests were performed on a Mettler
Table 1. TMA 40 thermomechanical analyzer coupled to the TC 10 A
The samples were obtained from the bulk material of the microprocessor and the printer above described. In this kind
ceramic object. To avoid surface contamination a very small of analysis the powdered samples were placed in cylindrical
point of the object was “cleaned” with a surgery scalpel. A alumina sample-holders (5 mm in diameter and 5 mm high)
slow battery electric drill was used to obtain about 600– equipped with an alumina piston capable of sliding inside the
800 mg of bulk (removing the first millimetres of obtained cylindrical sample holder and in close contact with the
levelled out surface of the sample. All the samples were
subjected to an isothermal (25 °C) recompaction process
Table 1 repeated three times [11,13]. This entailed applying a constant
History, provenience and classification of studied ancient pottery finds
load of 0.4 N for 10 min on the piston together with a
Samples Group Mark dynamic charge of 0.1 N (at a frequency of 5 cpm). At the
Rome Renaissance pottery 1 Rome Renaissance pottery (CeR) CeR1 end of this treatment the sample was subjected to thermo-
Rome Renaissance pottery 2 CeR2 dilatometric scanning between 25 °C and 1000 °C, in the
Rome Renaissance pottery 3 CeR3
same cylinder as described, at a heating rate of 8 °C min− 1,
Rome Renaissance pottery 4 CeR4
Rome Renaissance pottery 5 CeR5 in static air conditions and with a constant applied load of
Acephalous statue 5601 Civitella di Chieti (TeC) 5601 0.05 N.
Acephalous statue TestCol The TMA plots thus obtained are very similar to those found
Fictile statue B 19990B Civitella di Chieti (FrB) 19990B by Bayer and Wiedermann [14,15] using the same technique,
Fictile statue B 5597B 5597B
although this time on pottery samples that had not previously
Fictile statue B 19974B 19974B
Fictile statue A 5605A Civitella di Chieti (FrA) 5605A been powdered. This supports the belief that the recompacting
Fictile statue A 19973A 19973A treatment described here and used by us was largely successful
Fictile statue A 5602A 5602A in achieving the stated aim.
Ariccia statue 1 Ariccia (ArI) AR1 All the data were obtained by ICP–AES, TC and TMA
Ariccia statue 2 AR2
analysis and elaborated by Lotus 123 v 9.8 (IBM Corporation,
Ariccia statue 5 AR5
Libyan Sahara find A Libyan Sahara (SaL) SLA Armonk, USA) to make data matrix, the var-var charts, the
Libyan Sahara find B SLB correlation matrix, and some data validation was obtained with
Libyan Sahara find C SLC Winidams (Unesco) and by MVSP (Kovach Computing,
Libyan Sahara find D SLD Anglesey, UK) to calculate Principal Components Analysis
Libyan Sahara find E SLE
(PCA) and charts design.

3. F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120 115
3. Results and discussion the median value, that was then utilized to represent the sample
[20].
By ICP–AES analysis we determined, for each sample, the The large difference between the numeric values and the
content in ppm of the 16 main elements (Table 2). The result different kinds of units of measure, corresponding to the values
concerning thermogravimetric data were exemplified collecting obtained by different analytical methods (mg, ppm, °C, %…),
them, for each thermogram, in two steps: needed a first treatment of the matrix, first of all for the values
obtained by ICP–AES only and then of a second treatment for
▪ Step 1: between 200 °C and 600 °C, constituted by different the whole matrix.
sub-steps, where the loss of mass is given to different The series of 22 elements obtained by ICP–AES was
contributions: decomposition of traces of organic matter and reported in ppm. A first observation of these data showed that
so called structural bonded water, i.e. small amount of water some species were not present in all the samples, or, anyway,
originating from the loss, under heating, of hydroxyl groups were present with trace values, comparable to the LOD value of
still present in some minerals (%A in Table 2) instrument. The ICP–AES data of these species, Be, Y, Ba, Sb,
▪ Step 2: between 600 °C and 750 °C, certainly due to the Sn, were so eliminated from the matrix, obtaining 16 elements
decomposition of carbonates (principally of Calcium) for 20 objects. In the group of the species that we removed the
present in the samples (%B in Table 2) most frequent was, anyway, the Sn, detected in only 8 samples,
with values included between 3 and 81 ppm and the least
Finally it was evaluated the value of residue too that, at frequent was the Be, detected in only one sample, with a value
1000 °C, is principally constituted by metal oxides and silicates of 0.3 ppm.
(%Res in Table 2). The thermogravimetric data, as above described, were
In addition, the equivalent firing temperatures, estimated on collected in two numeric values, corresponding to the mass
the basis of thermomechanical (TMA) curves [16] and, in percentages loss and to the data indicating the percentage
particular, the “shrinkage temperature” [17,18] obtained as well residue at the end of the TG analysis.
as described by Tite [18], was insert in Table 2 (fT). Adding the equivalent firing temperatures obtained by TMA,
One of the aims of a classification work is often to find that is the temperatures corresponding to the original firing
suitable descriptors that allow an application in other similar temperature, we obtained a 20 X 20 matrix (Table 2).
cases too. In the matrix the Zn was the specie with a largest number of
From this point of view it was therefore likely to consider gaps (5). The values of the present species were enclosed
that in this research work the study, as the potteries were between 10 and 95 ppm. In all, the whole matrix shows 15
originating from different sites and different periods, easily absences of data.
allowed a good “separation” of the different samples belonging The data fill was processed with a technique already used
to different archaeological sites and, therefore, an accurate by us with success in other cases: it provides for the
selection of the variables, even if using simple chemometric calculation of the lowest LOD value for the whole set of
methods. elements (all originating from the same instrument) and the
Naturally the 20 considered samples come from a more wide following refilling of the gaps with random values between 1/3
set and they were selected on the basis of the certainty of and 2/3 of the value before estimated. Alternatively, the data
geographical and hystorical origin. fill can be carried out with the calculation of the LOD value
It was used the most common analytical techniques also for each column and the filling of the gaps with the same
utilized in the literature for the analysis of pottery finds: ICP– method.
AES spectrometry, Thermogravimetric analysis, Thermome- That allowed to not eliminate rows and/or columns if the data
chanical analysis. fill concerned a small number of values, and to not insert zero
X-ray diffractometric analysis [19] was also carried out for values that, for similar object, were present in significant
the powders, but the corresponding data were found definitely amount, so that very small values (but measurable) contributed
redundant, so they are not inserted in the Table 2. to the classification.
Using the techniques above we obtained, for each sample, a From this point of view, the “zero” value was assigned to the
matrix constituted by the content of the 22 chemical species by elements “certainly not present” and not to the elements present
ICP–AES, by the value of equivalent firing temperature by with values of concentration lower than the limit of detection
TMA, and the percentage of loss on mass % for the first and and it was able too to exert a role in the classification.
second step and by the percentage of residual at 1000 °C by At this point the obtained matrix (20 X 20) showed average
TGA. values very different. The main components of ceramics Si, Al,
For the multivariate analysis we selected the PCA, as it Ca, Fe showed average values between 5 and 70%. The
allows a rapid visual monitoring of obtained results [9], an easy intermediate components Mg, Ti, As showed values below 5%.
interpretation of new variables and also a study of the loadings, The trace components Zn, Sr, Cu, Mn, Zr, Os, Cr, La, Pb
that is precisely one of the main target of the work. showed values lower than 1%.
Each sample was split in three smaller fragments and each To allow to all the components of the matrix to reveal their
of them was independently analysed by the different discriminant power, the matrix was “studentised” (i.e. each
techniques. By three values so obtained each time we evaluated value in column was subtracted from the arithmetic average,

4. 116 F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120
Table 2
The matrix used for multivariate analysis, after autoscaling
Sample Group Si Al Ca Fe Mg Ti As Zn Sr
CR1 CeR − 0.9560 0.7119 0.8811 0.2267 − 0.8787 0.0643 − 0.1032 −1.2520 −0.5538
CR2 CeR − 0.9453 − 0.2406 1.3208 − 0.2990 0.5838 − 0.7763 − 0.3226 0.5546 1.5174
CR3 CeR − 0.7260 − 0.5385 1.3990 − 1.0926 0.1891 − 0.8514 − 0.2877 −0.0490 1.3654
CR4 CeR − 1.7741 0.1661 2.0071 0.2183 0.7033 − 0.5260 0.5514 0.2403 3.0569
5601 TeC − 0.6089 − 0.3867 0.8620 0.3295 0.6886 − 0.9352 − 0.2833 1.0483 0.2431
TestCol TeC − 0.7724 − 0.8574 1.3511 − 0.1681 1.2147 − 0.8431 0.6696 −0.5408 −0.5731
19990B FrB − 0.2396 0.0110 0.1285 0.2618 1.2770 − 0.6229 − 0.1886 −1.2552 0.2078
5597B FrB − 0.0847 0.0808 − 0.2277 1.0183 0.9035 − 0.1152 0.0697 −0.3339 0.0508
19974B FrB − 0.0395 − 0.1225 0.1000 − 0.1568 0.7236 − 0.8032 0.2019 −1.2552 0.0046
5605A FrA 0.6949 0.9952 − 0.9391 − 1.6114 − 1.3867 − 0.7112 0.2343 1.8550 −1.2201
19973A FrA − 0.2365 1.9194 − 0.8246 − 0.0162 0.3594 0.7866 − 1.0118 −0.0003 −0.1132
5602A FrA 0.2909 1.2388 − 0.8494 − 0.5440 − 0.8164 0.0508 − 0.3271 0.8069 −0.9641
AR1 ArI − 0.6349 0.2738 0.2767 1.4718 0.8577 − 0.0417 1.3157 −1.2520 0.1350
AR2 ArI 0.0320 0.1104 − 0.4775 1.4809 0.8993 0.1498 1.1205 −1.2552 0.0090
AR5 ArI − 0.9248 1.2125 0.0937 1.6440 0.7645 0.2479 2.7462 −0.9143 0.0379
SLA SaL 1.3930 − 1.6197 − 0.9311 1.2864 − 1.4610 0.7168 − 1.0784 0.3401 −0.5280
SLB SaL 1.2228 − 0.1912 − 1.1723 − 0.6279 − 1.4324 0.9745 − 0.5548 0.5300 −1.0649
SLC SaL 1.4532 − 1.0839 − 1.0832 − 0.6998 − 1.1474 3.4498 − 1.2388 1.2350 −0.4453
SLD SaL 0.7296 0.6096 − 0.9311 − 1.0840 − 1.0470 0.1752 0.2497 −0.0562 −0.6437
SLE SaL 2.1264 − 2.2889 − 0.9842 − 1.6378 − 0.9950 − 0.3897 − 1.7627 1.5540 −0.5216
obtained only using values different from zero, and divided by matrix we can see all var-var correlation plots and look for linear
standard deviation of the sample). correlation (available also in many software only as text matrix).
It was not omitted even the calculation and the representation In the figure we could see the hard correlation between the
of the matrix of correlation scatter plots, shown in Fig. 1. In this second % loss and the % residue for TG analysis (%B and RES
Fig. 1. Correlation among all the 20 variables, raw data. In the diagonal the form of the distribution of the variables.

5. F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120 117
Cu Mn Zr Os Cr La Pb firT − %A − %B Res%
− 1.3598 − 1.0104 0.7882 −0.0512 −1.1023 1.8561 0.0007 0.7891 0.8967 0.6212 1.4226
− 0.6638 − 0.7385 − 0.7775 −0.1736 −0.2749 0.1618 − 0.0834 0.8732 1.0904 − 1.7028 − 1.8593
− 0.6671 − 0.8163 − 0.3023 1.9888 −0.9077 1.0261 − 0.5445 0.8900 1.0904 0.6956 1.4743
− 0.5038 1.0918 − 1.1586 1.3214 −0.2275 0.6795 − 0.5445 0.7386 1.0904 0.9187 0.7766
− 0.6168 − 1.2000 − 0.7541 0.4290 −0.9444 0.6319 − 0.5228 0.6208 1.0904 − 1.1166 − 0.1538
− 0.4607 − 0.8592 − 0.3973 0.7915 −0.2898 − 0.1237 − 0.4707 0.4189 − 0.3725 − 0.5602 − 0.5414
− 0.3774 − 1.2000 − 1.1512 0.4508 2.0089 − 0.0672 − 0.5228 0.5872 1.0904 − 0.1346 − 1.1099
0.7169 2.0771 − 0.5057 −0.0775 1.3279 − 1.0533 − 0.1558 0.6040 1.0904 − 0.8458 − 0.5931
− 0.2466 − 1.1931 − 1.1512 0.4001 −0.9416 0.2498 − 0.4635 0.8227 1.0904 − 1.4826 − 1.2133
− 1.1598 − 0.7042 − 0.0544 −0.9620 −0.1705 − 0.6857 2.7161 0.7891 − 0.7504 − 1.0601 − 1.1099
0.5635 1.0275 2.2029 −0.8526 0.7722 − 0.6539 2.6590 0.6881 − 0.5566 − 1.5659 − 1.3942
0.0867 − 0.1589 1.3621 −0.8613 0.0393 − 0.7626 1.2412 0.7386 − 1.3317 − 0.3459 − 0.2571
0.7375 0.9044 0.0012 0.6844 2.5824 1.1710 − 0.4695 − 0.1531 − 0.2660 0.0409 0.0788
0.4164 1.1978 − 0.4408 1.4914 −1.0347 1.8619 − 0.4761 − 0.7588 − 1.3317 0.9931 0.6215
1.3107 0.6824 − 0.0600 0.9438 0.3289 0.8726 − 0.4287 0.0320 − 1.5255 0.6658 0.5698
− 0.6254 1.1158 − 0.5952 −1.1245 −0.1294 − 1.0073 − 0.2824 − 1.4486 − 0.7698 0.9068 − 0.1796
1.4370 − 0.7118 1.5704 −1.0353 −0.1939 − 1.0799 − 0.4982 − 1.5159 0.1603 0.9931 1.0350
2.6611 − 0.0796 0.1694 −1.1073 −0.2301 − 1.0202 − 0.4002 − 1.1458 − 1.1379 0.9931 0.3631
− 0.2512 0.2684 1.6670 −1.1109 −0.7549 − 0.9663 − 0.4061 − 1.7346 0.3057 0.9931 1.1642
− 0.9977 0.3067 − 0.4129 −1.1453 0.1421 − 1.0905 − 0.3478 − 1.8356 − 0.9538 0.9931 0.9058
%) and for the selection of the variables we retained the first and terracotta samples, near to the other values equal to 0.72 and
the second step. 0.68. Unfortunately the autoscaling process emphasised the
We also focalised the attention on the content of Mg, which differences and the objects were retained.
contributed to the separation between two groups of the values, A first calculation, using all the variables, showed values of
therefore Mg was another possible candidate for variable variance equal to 34.4%, 51.4% and 61.2% for the first three
discharge. components.
The content of Ti showed an outlier in this graphs (the shown The study of the loading highlighted vectors with similar
values were autoscaled). direction and size, allowing the elimination of “Mg” and “R%”,
Looking for the raw data we found a value equal to 1. 1% for already above described, and of the species “Cr” and “Al”
the Ti species in the case of the sample C of prehistoric Libyan because of their small contribution to the first two components.
Fig. 2. Scores chart, after autoscaling. 40.7% and 22.7% of variance. Data linked by provenience.

6. 118 F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120
Table 3 the Fe, it was observed that the two samples AR2 and AR (II–III
Eigenvalues computed on autoscaled data using 11 variables century A.D.) presented the highest amount of it (maximum
Number of Eigenvalues Retain. Cum. retain. RSS Press Press/ values of PC2) and the sample 5605A (I century) the lowest
PCs var. % var. % RSS amount.
#1 4.5 40.7 40.7 123.9 142.5 1.2 The graphical representation highlighted moreover that the
#2 2.5 22.7 63.4 76.4 106.0 1.4 two samples defined as unknown, but with the same style of
#3 1.5 13.2 76.7 48.7 76.7 1.6
“frontonal” finds (I–II century A.D.), showed values of PC1
#4 0.9 7.8 84.5 32.4 62.5 1.9
#5 0.7 6.1 90.6 19.7 41.8 2.1 and PC2 suggesting a composition similar to the group
#6 0.4 3.4 94.0 12.5 31.0 2.5 “frontonale B”. PC1 vs PC3 chart showed similarly the two
unknown objects “near” the two “frontonal B” (1990b and
19974b).
By means of a further selection we lastly retained the species It was clear that at least one of them (“testa colossale”) could
Si, Ca and Fe, as “major chemical elements”, Ti as low value not belonging to “frontonale B” because of their large different
chemical elements and Zn, Cu, La, Pb as traces, with estimated sizes; however its chemical and firing characteristics were very
equivalent firing temperature and the first and second thermo- similar to those of the finds belonging to the “frontone” B.
gravimetric step. Having at this point a series of variables originating from the
Using these 11 variables the 20 objects were well separated three analytical techniques, which showed a high discriminant
in the only two PC (with cumulative variance PC1 = 40.7, power with a high selectivity, that is a pattern able to provide a
PC2 = 63.4, PC3 = 76.7, PC4 = 84.5%). In Fig. 2 it is shown the good separation of the groups, it was possible to study other
projection over the first two components and in Table 3 the analytical points.
obtained values. As it was possible to find works where the clustering was
Looking for the data and groups we found the PC1 as “time obtained starting from the data coming from a single analytical
axis”, that is the first component linked to the dating of the technique, by means of this data set it was possible to try to
sample, therefore with the more ancient finds of Libyan analyse the results coming by a single method.
samples, with low values of PC1 and the renaissance ceramics, In Fig. 3 it is shown the result obtained for the first two PC
with high values of PC1, proceeding through the roman period (with a variance PC1 = 38.7%, PC2 = 56.3%, PC3 = 69.2%
finds Single inversion of the time axis was present by the three respectively) using all the 16 elements. It is noticeable a
fictile statues belonging to the pediment, defined as “type A”, “separation”, but the selectivity is partial and, practically, only
probably with different origin. the Ariccia ceramics constitute a separate group, confusing also
The PC2 looked linked with the presence of minor the time scale.
components and, in particular, with the species Pb, Zn, and Alternatively we tried to using only the values of elements
Fe, that worked as “separators” within the groups; concerning before selected. The values of the 8 species provided a best
Fig. 3. Scores chart, after autoscaling, using all the 16 elements measured by ICP. Variance 38.7% and 17.7%.

7. F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120 119
Fig. 4. Scores chart, after autoscaling, using only the 8 elements (measured by ICP) obtained with variables selection. Variance 45.1% and 25.3%.
classification (see Fig. 4), but, anyway, worse than that obtained and of the TMA combined with other analytical methods for the
with the variables belonging from the three analytical studies of pottery finds.
techniques, even if showing values similar for the variance The Fig. 5 shows the loading of 11 variables selected and it
explained by the first three components (PC1 = 45.1%, highlights how the third quadrant was filled only by the first
PC2 = 70.4%, PC3 = 81.2%). percentage of loss of mass and by the equivalent firing
With only two variables, belonging from TG, combined with temperature, providing a high contribute to the classification.
the equivalent firing temperature, we obtained a high dis- The figure also shows the application of the variables
criminant power. This fact suggested the application of the TG selecting method, often using by us: we retain the variables that
Fig. 5. Loading chart, autoscaling, all shown but only the circle blue variable are retained after variables selection. (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this article.)

8. 120 F. Bellanti et al. / Microchemical Journal 88 (2008) 113–120
provide the greater contribute to the axes on which we want to di Chieti: il frontone delle muse, on Deliciae Fictiles III. Architectural
work; if we choose, for example, the plane PC1/PC2 we try to Terracottas in Ancient Italy: New Discoveries and Interpretations, in: G.
Greco, J. Kenfield, Edlund-Berry (Eds.), Proc. of the International
retain the loadings that show the greater value for PC1 or PC2, Conference Held at the American Academy in Rome, 7-8/11/2002, 2006.
leaving out the loadings with the values close to zero. In figure [6] E. Marengo, M. Aceto, E. Robotti, M.C. Liparota, M. Bobba, G. Pantò,
the green square points show variables not used for PCA Archaeometric characterisation of ancient pottery belonging to the
calculation. archaeological site of Novalesa Abbey (Piedmont, Italy) by ICP–MS
and spectroscopic techniques coupled to multivariate statistical tools, Anal.
Chim. Acta 537 (1-2) (2005) 359–375.
4. Conclusions [7] D.N. Papadopoulou, G.A. Zachariadis, A.N. Anthemidis, N.C. Tsirliganis,
J.A. Stratis, Microwave-assisted versus conventional decomposition
The use of three analytical instrumental methods, the first procedures applied to a ceramic potsherd standard reference material by
(ICP–AES) that determined the percentage content of chemical inductively coupled plasma atomic emission spectrometry, Anal. Chim.
species mainly present or present in traces, a second that was Acta 505 (1) (2004) 173–181.
[8] C. Rathossi, P. Tsolis Katagas, C. Katagas, Technology and composition of
able to estimate the firing temperature and the third (TG) that Roman pottery in northwestern Peloponnese, Greece, Appl. Clay Sci. 24
was able to evaluate the two main losses of mass and the (3-4) (2004) 313–326.
residual mass at 1000 °C, allowed a good separation in the space [9] RQ-mode principal components analysis of ceramic compositional data,
of the first principal components. An accurate selection of the Archaeometry 36 (1994) 115–130.
variables improved the selectivity and highlighted the con- [10] A. Moropoulou, A. Bakolas, K. Bisbikou, Thermal analysis as a method of
characterizing ancient ceramic technologies, Thermochim. Acta 269-270
tribute of the TG in the analysis of pottery. (1995) 743–753.
The good result obtained attests also the use of a mix of data [11] M. Tomassetti, L. Campanella, P. Flamini, G. Bandini, Thermal analysis of
corresponding to the species present as components: principal, fictile votive statues of 3rd century B.C. Thermochim. Acta 291 (1-2)
minor, or in traces, as good descriptor of the objects. With (1997) 117–130.
regard to it our group is studying the same kind of research [12] H. Neff, M.D. Glascock, R.L. Bishop, M.J. Blackman, An assessment of
the acid-extraction approach to compositional characterization of archae-
based on similar approach, that shows how this strategy is ological ceramics, Am. Antiq. 61 (2) (1996) 389–404.
useful in the classification of other kind of finds too, such as [13] L. Campanella, G. Favero, P. Flamini, M. Tomassetti, Prehistoric
marbles and ancient glasswares. terracottas from the libyan tadrart acacus, J. Therm. Anal. Calorim. 73
(1) (2003) 127–142.
References [14] H.G. Wiedemann, G. Bayer, Approach to ancient Chinese artifacts by
means of thermal analysis, Thermochim. Acta 200 (1992) 215–255.
[15] G. Bayer, H.G. Wiedemann, Thermoanalytical measurements in archaeo-
[1] G. Olcese, La produzione ceramica ad "Albintimilium" (Liguria) in epoca metry, Thermochim. Acta 69 (1-2) (1983) 167–173.
tardo romana: analisi chimiche e mineralogiche, on L. Paroli, La ceramica [16] A.B. Searle, R.W. Grimshaw, The Chemistry and Physics of Clays and
invetriata tardoantico e altomedievale in Italia, all'Insegna del Giglio Ed,
other Ceramic Materials, Interscience Ed., New York, ISBN 0510477011,
Florence, Italy, 1992, pp. 621–636. 1959.
[2] S. Polla, M. de Vos, P. Fermo, S. Gialanella, G. Ischia, Studio [17] J.P. Roberts, Determination of firing temperature of ancient ceramics by
comparativo di manufatti ceramici di produzione africana, Proc. of III
measurement of thermal expansion, Archaeometry 6 (1963) 20–25.
Congresso Nazionale AIAR, Bressanone, Italy, Cleup Ed., Padova, 2004,
[18] M.S. Tite, Determination of the firing temperature of ancient ceramics by
pp. 118–119. measurement of thermal expansion: a reassessment, Archaeometry 11
[3] M.S. Tite, Pottery production, distribution, and consumption — the (1969) 131–143.
contribution of the physical sciences, J. Archaeol. Method Thoery 6 (3)
[19] J. Zhua, J. Shan, P. Qiu, Y. Qin, C. Wang, D. He, B. Sun, P. Tong, S. Wu,
(1999) 181–233. The multivariate statistical analysis and XRD analysis of pottery at
[4] M. Cremaschi, S. Di Lernia, E.A. Garcea, Some insights on the Aterian in Xigongqiao site, J. Archaeol. Sci. 31 (12) (2004) 1685–1691.
the Libyan Sahara: chronology environment and archaeology, Afr. [20] J.M. Cogswell, H. Neff, M.D. Glascock, Analysis of shell-tempered
Archaeol. Rev. 15 (4) (1998) 261–286.
pottery replicates: implications for provenance studies, Am. Antiq. 63 (1)
[5] G. Iaculli, Note sulla tecnica di esecuzione di alcune terrecotte della civ- (1998) 63–72.
itella di Chieti and D. Liberatore, Le terrecotte architettoniche delle civitella