Information Technology Reference
In-Depth Information
6.1.5.3 Ceramic Physical Properties and Ultrasound Propagation
The differences in physical properties, composition and processing of the ceramic
pieces, presented above, suggest the possibility of devising non-destructive tech-
niques for archaeological ceramic classification. In Sect. 6.1.5.1 was shown that
the pieces could be separated by chronological periods using measures of their
open porosity and apparent density. Besides, it is well-known that porosity and
density of a material have a definite influence on the propagation of the ultrasound
[ 22 , 23 ]. Thus, it is clear that should be there correlation between the results
obtained by the proposed method based on ultrasounds ( Sect. 6.1.4 ) and the dif-
ferences in physical properties of the pieces for the different chronological periods.
There are several factors that can determine the porosity and density of
ceramics, such as the raw material composition and the processing method
employed to manufacture the pieces. However, in the case of archaeological
ceramics, the original ceramic physical properties after manufacturing, can be
altered by other factors such as the ceramic use (i.e., over-heating for cooking,
etc.) and in general with the pass of the time (i.e., fractures, loss of cover layers,
etc.). Thus, an exhaustive analysis of physical properties and how these properties
were derived for archaeological ceramics becomes a very complex problem that
needed an important amount of information that is outside the scope of this work.
Note that the objective of this work is to provide a new NDT procedure to classify
archaeological ceramics from the basis of training with a set of pieces of known
class made with the intervention of an expert. A correct training will determine the
achievement of the procedure to classify ceramics of unknown class.
The analysis of the results obtained by ultrasounds provided here consider
correct (or at least probabilistic) labelling made by the expert and are based on
available data of the composition, processing and physical features of the ceramics
shown in Sects. 6.1.5.1 and 6.1.5.2 . Let us explain the misclassifications in the
confusion matrix of the ultrasound-based classification of Table 6.3 . Misclassifi-
cation is obtained from similar responses of pieces from different periods to the
ultrasounds. Table 6.3 shows that Roman ceramics is the most misclassified group.
Confusion between Roman and Iberian pieces (19 and 9 %) can be explained from
ceramic composition and processing. The amphora and common Roman pieces
were made from iron-rich paste and using lathe as well as the Iberian pieces. Thus,
the mechanical and physical properties for these two groups were similar.
The confusion between Roman and Bronze Age pieces (5 and 7 %) can be
explained due to changes in the structure of some of the Roman pieces of the
sigillata subgroup that had lost the cover varnish. The high value of porosity
shown by the fragments of sigillata is associated with pores of very small size and
very connected, which allows big water absorption once the varnish is removed.
Thus, these two groups of pieces show similar physical properties due accidents
cause with the pass of the time. Regarding to the confusion between Bronze Age
and Middle Ages pieces (14 and 2 %), this also can be explained from composition
and processing. The Islamic subgroup of Middle Ages pieces were from the
''paleoandalusí'' period (early centuries of the Islamic period in Spain). During,
Search WWH ::




Custom Search