Glass and Ceramics 163
when reacting with silica to form a glass: the silicon–oxygen network is
broken up (Figure 3). The ionic bonds between these network modifier cations
and non-bridging oxygen anions from the network are less strong than the links
within the network. Therefore, cations, such as potassium and calcium ions,
are able to migrate within the structure, leading to a lower chemical durability.
Furthermore, there is a third group of oxides (e.g.of lead, aluminium, mag-
nesium) that can be incorporated into the glass structure in both ways, as
network modifier or as network former.
Sand (SiO 2 ) was the main component for the production of ancient glass.
Lime (CaCO 3 ) or magnesium carbonate (MgCO 3 ) and soda (Na 2 CO 3 ) or
potash (K 2 CO 3 ) were added to reduce the melting temperature and to facilitate
the production process. For soda- or potash-lime silicate glasses, these elements
represent more than 90 wt% within the complex composition. Nevertheless,
most historic glasses contain up to 30 different components that are present
only in minor quantities or as trace elements. Some of them have not been
added on purpose, but are present as impurities contained in the main com-
ponents. Others have been chosen as colouring agents. It should be stressedthat
the colour of a glass is determined not only by the amount of a specific tran-
sition metal oxide but also by its oxidation state, which can be regulated during
the production process by controlling the atmosphere in the furnace. Non-
transparent but opaque glasses can be obtained by the presence of opacifiers,
which are mainly tin, lead or antimony compounds.
In spite of some general rules for dating glasses by their chemical compos-
ition, there are many uncertainties due to the fact that the glass production of
each period exhibited a broad variety in composition, although some recipes
remained the same over the centuries. Nevertheless, archaeometric studies can
Figure 3Scheme of the structure of a potash-lime silicate glass. (Note: the fourth bond of
silicon is out of the plane and not represented in the two-dimensional scheme)
(After Scholze, 1988.)