Sandra Davison.Conservation and Restoration of Glass
Contents
About the author
vi
Preface
vii
Acknowledgements
ix
Introduction
xi
1The nature of glass
1
2Historical development of glass
16
3Technology of glass production
73Part 1: Methods and materials73Part 2: Furnaces and meltingtechniques135
4Deterioration of glass
169
5Materials used for glassrestoration
199
6Examination of glass, recording and documentation
227
7Conservation and restoration of glass
242Part 1: Excavated glass243Part 2: Historic and decorative glass271
Appendix 1 Materials and equipment for glass conservation and restoration
345
Appendix 2 Sources of information
347
Bibliography
349
Index
367
About the author
Sandra Davison FICC ACR trained in archaeo-logical conservation at the Institute of Archaeology (London University), and has worked as a practising conservator for thirty-five years. Fourteen years were spent as aconservator at The British Museum, and aftera brief spell abroad, she has continued in herown private practice since 1984. Sandra haslectured and published widely, including adefinitive work,
Conservation of Glass
(withProfessor Roy Newton, OBE), of which this volume is a revised and enlarged edition.In addition to working for museums in theUnited Kingdom, France, the Czech Republic,Malaysia and Saudi Arabia, she has taughtglass restoration in the UK, Denmark, Norway,the Netherlands, the USA, Egypt, Mexico and Yugoslavia.In 1979 she was made a Fellow of theInternational Association for the Conservationof Historic and Artistic Works (IIC), and in2000 became one of the first conservators tobecome an accredited member of the UnitedKingdom Institute for Conservation (UKIC).
Preface
Conservation of Glass , first published in 1989, was intended to serve as a textbook for conser- vation students, conservators and restorers working on glass artefacts within museums,and those restoring painted (stained) glass windowsin situ . It was written by two authors with very different, but complementary backgrounds and experience in the conserva-tion of glass. Roy Newton, a glass scientist(now retired), has worked in glass manufac-turing, on the archaeology of glass and on theproblems concerned with the conservation of medieval ecclesiastical painted windows.Sandra Davison, a practising conservator forover thirty years, has conserved a great variety of glass artefacts, published and lectured widely, and teaches the principles and practiceof glass conservation in many countries.In this edition, written by Sandra Davison,the section concerning painted glass windowrestoration has been removed, with the inten-tion of producing a separate volume at a laterdate. However, information concerning thehistory and technology of glass window-making has been retained as backgroundknowledge for conservators preserving panelsof glass held in collections. The revised title,Conservationand Restoration of Glass , reflectsthe closer involvement of conservators indeveloping conservation strategies for dealing with glass in historic houses and elsewhere inthe public arena. The volume includes sectionson the historical development and treatmentof mirrors, chandeliers, reverse paintings onglass and enamels.Conservation and Restoration of Glass provides an introduction to the considerablebackground knowledge required by conserva-tors and restorers concerning the objects intheir care. Chapter 1 defines the nature of glass in terms of its chemical structure andphysical properties. Chapter 2 contains a brief history of glassmaking, illustrating the chang-ing styles of glass decoration, and the histori-cal development of light fittings (in particularchandeliers), flat glass, mirrors, reverse glasspaintings and micromosaics and enamels.Chapter 3 consists of two parts. The firstdescribes the use of the raw materials from which glass is made and the historical devel-opment of methods of glass manufacture; thesecond is concerned with the development of furnaces and melting techniques. The mecha-nisms by which glass deteriorates, in differentenvironments, are described in Chapter 4,together with an outline of experiments under-taken for commercial/industrial concerns, todetermine the durability of glass. The materi-als used in the processes of conservation andrestoration of glass are discussed in Chapter 5.The examination of glass, described in Chapter6, outlines both simple methods for use by conservators, and those more elaboratetechniques which can be of use for analysis,research and the detection of fakes. Finally, in Chapter 7, the details of conservation andrestoration techniques, based on currentpractice in several countries, are described andillustrated. Conservators/restorers should notnormally undertake complicated proceduresfor which they have not had training orexperience; but specialized areas of glassconservation are outlined in Chapter 7 in orderto identify the problems that will requireexpert attention. Information concerningdevelopments in glass conservation, whichmay also include details of treatments thathave proved to be unsuccessful, can be foundin conservation literature and glass conferenceproceedings.
Acknowledgements
There have been significant developments andgrowth in glass conservation. The author hasattempted to reflect this by inviting commentsfrom a number of conservators and restorers(in private practice or museum employment),conservation scientists and experts in relatedfields, working in Britain, Europe and North America.In particular, the author is greatly indebtedto Professor Roy Newton for undertaking theenormous amount of research for Conserva- tion of Glass , of which this book is a devel-opment; and to the following colleagues fortheir valuable assistance (and who, unlessstated otherwise, are in private practice):Chapter 1: Angela Seddon (Professor of Materials Science, University of Nottingham).Chapter 2: Phil Barnes (enamels); Simone Bretz(reverse paintings on glass; Germany); Judy Rudoe (micromosaics; Assistant Keeper,Department of Medieval and Modern Europe,British Museum); Mark Bamborough (paintedglass windows); Tom Kupper (plain glazing;Lincoln Cathedral); Eva Rydlova (Brychta glassfigurines; Czech Republic). Chapter 3 part 1:Paul Nicholson (Egyptologist, University of Bristol); part 2: David Crossley (industrialarchaeologist, The University of Sheffield) andthe late Robert Charleston (glass historian andformer Curator of the Department of Ceramicsand Glass, Victoria and Albert Museum).Chapter 4: Ian Freestone (Deputy Keeper,Department of Scientific Research, BritishMuseum). Chapter 5: Velson Horie (conserva-tion scientist, Manchester Museum, University of Manchester). Chapter 6: Angela Seddon(University of Nottingham) and Ian Freestone(British Museum). Chapter 7: Victoria Oakley (Head of Ceramics and Glass Conservation, Victoria and Albert Museum) and Patricia Jackson (UK), Rolf Wihr (Germany), CarolaBohm (Sweden), Raymond Errett (retired) andSharon Smith-Abbott (USA) (glass object conser- vators); Alison Rae and Jenny Potter (conserva-tors of ethnographic material – beads; Organic Artefacts Section, Department of Conservation,British Museum); Annie Lord (textile conserva-tor – beads; The Conservation Centre, NationalMuseums and Galleries Merseyside, Liverpool).Thanks are also due to Vantico (formerly Ciba Speciality Polymers), Duxford, Cambridgefor technical advice and for a generous granttowards research. Finally to my family, T.K.and E. Lord, without whose gift of a computerthis book would not have been written, to WBJH for patience with computer queries andendless photocopying, and Steve Bell fortechnical support.The sources of illustrations (other than thoseby Roy Newton and the author) are statedbriefly in the captions. Every effort has beenmade to trace copyright holders. The authorand publishers gratefully acknowledge thekind permission, granted by individuals,museum authorities, publishers and others, toreproduce copyright material.
S.D.2002
Introduction
The conservation of glass, as of all artefacts,falls into two main categories:
passive conser- vation , the control of the surrounding environ-ment to prevent further deterioration; and active conservation , the treatment of artefactsto stabilize them. A storage or display environ-ment will consist of one of the following: (i)natural climatic conditions (especially paintedglass windows and glass mosaics in situ ); (ii)modified (buffered) climatic conditions inbuildings and cases with no air conditioning;(iii) controlled climatic conditions, where airconditioning has been installed in museumgalleries or individual showcases, to holdtemperature and relative humidity withincarefully defined parameters. Environmentalcontrol is a discipline in its own right(Thomson, 1998) and outside the scope of thisbook. However, conservators need to beaware of the basic facts in order to be able toengage in discussions regarding display andstorage conditions, and the choice of materi-als for display, and packaging for storage andtransport. The prevention of further damageand decay by passive conservation , representsthe minimum type of treatment, and normally follows examination and recording. Reasonsfor not undertaking further conservation mightbe lack of finance, facilities, lack of an appro-priate treatment or the sheer volume of glass,e.g. from excavation.
Active conservation , as the term implies,involves various levels of interference.
Minimal conservation would include ‘first aid’,photography, X-radiography (where appropri-ate), a minimal amount of investigative conser- vation such as surface cleaning, and suitablepackaging or repackaging for safe storage.
Partial conservation entails the work abovebut with a higher degree of cleaning, with or without consolidation.
Full conservation work would additionally involve consolidation andrepair (reconstruction of existing fragments),supplemented by additional analytical infor-mation where appropriate.
Display standard conservation might include cosmetic treatmentsuch as restoration (partial or full replacementof missing parts) or interpretative mounting fordisplay. Restoration of glass objects may alsobe necessary to enable them to be handledsafely. It should only be carried out accordingto sound archaeological or historical evidence.The level of conservation has to be agreedbetween a conservator/restorer and the owner,custodian or curator, before work begins.Historically, glass conservation was not aseasily developed as it was for ceramics, forexample. The fragile nature of glass made itdifficult to retrieve from excavations, and thetransparent quality of much glass posed thedifficulty of finding suitable adhesives andgap-filling materials with which to work. Theuse of synthetic materials and improvementsin terrestrial and underwater archaeologicalexcavation techniques have resulted in thepreservation of glass which it was not formerly possible to retrieve; and continues to extendthe knowledge of ancient glass history,technology and trade routes. Early treatmentsusing shellac, waxes and plaster of Paris wereopaque or coloured and not aesthetically pleasing (Davison, 1984). Later, rigid transpar-ent acrylic materials such as Perspex (US:Plexiglas) were heat-formed and cut to replacemissing areas of glass. Advantages were theirtransparency and only slight discoloration andembrittlement with age. However, theprocesses were time-consuming, and thereplacements did not necessarily fit wellagainst the original glass. Unweathered glass surfaces are smooth, essentially non-porousand are covered with a microscopic layer of water, so that few materials will adhere satis-factorily to them. It was only with thecommercial formulation of clear, cold-settingsynthetic materials, with greater adhesiveproperties, that significant developments inglass conservation were achieved. Epoxy,polyester and acrylic resins could be polymer-ized in moulds in situ , at ambient tempera-tures with little or no shrinkage. However,restoration involves interference with the glassin terms of the moulding and casting processes(Newton and Davison, 1989). Recentapproaches to glass conservation and restora-tion have been the construction of detachablegap-fills (Hogan, 1993; Koob, 2000), and themounting of glass fragments or incompleteobjects on modern blown glass formers, or onacrylic mounts.
1 The nature of glass
The term glass is commonly applied to thetransparent, brittle material used to form windows, vessels and many other objects.More correctly, glass refers to a state of matter with a disordered chemical structure, i.e. non-crystalline. A wide variety of such glasses isknown, both inorganic (for instancecompound glasses and enamels, and even thesomewhat rare metallic glasses) and organic(such as barley sugar); this book is concernedonly with inorganic glasses, and then only with certain silicate glasses, which areinorganic products of fusion, cooled to a rigidcondition without crystallizing. The term ancient glasses is that used by Turner(1956a,b) to define silicate glasses which weremade before there was a reasonable under-standing of glass compositions, that is beforethe middle of the seventeenth century (seealso Brill, 1962). In this book, for convenience,the term glass will be used to mean bothancient and historic silicate glasses.Understanding the special chemical structureand unique physical properties of silicateglasses is essential in order to appreciate boththe processes of manufacture of glass objectsand the deterioration of glass, which may make conservation a necessity.
Natural glasses
Before the discovery of how glass could bemanufactured from its raw ingredients, manhad used naturally occurring glass for many thousands of years. Natural silica (the basicingredient of glass) is found in three crystallineforms, quartz, tridymite and cristobalite, andeach of these can also occur in at least two forms. Quartz is the most common, in theform of rock crystal, sand, or as a constituentof clay. Rock crystal was fashioned into beadsand other decorative objects, including, inseventeenth century France, chandelier drops.If quartz is free from inclusions, it can be visually mistaken for glass.Sudden volcanic eruptions, followed by rapid cooling, can cause highly siliceous lavato form natural glasses (amorphous silica), of which obsidian is the most common. Inancient times, obsidian was chipped andflaked to form sharp-edged tools, in the samemanner as flint (Figure 1.1). Other forms of naturally occurring glass are volcanic pumice,lechatelierite or fulgurites and tektites. Pumiceis a natural foamed glass produced by gasesbeing liberated from solution in molten lava,before and after rapid cooling. Lechatelierite isa fused silica glass formed in desert areas by lightning striking a mass of sand. The irregu-lar tubes of fused silica (fulgurites) may be of considerable length. Lechatelierite has alsobeen discovered in association with meteoritecraters, for example at Winslow, Arizona.Tektites are small rounded pieces of glass, of meteoric origin, found just below the surfaceof the ground in many parts of the world, and which appear to have come through theatmosphere and been heated by fallingthrough the air while rotating. Their composi-tion is similar to that of obsidian, but they contain more iron and manganese.
Man-made glasses
In order to understand the nature of man-made glass, it is first necessary to defineseveral terms for vitreous materials, some of which have previously been used ambiguously or incorrectly (Tite and Bimson, 1987). Thereare four vitreous products: glass, glaze, enameland (so-called, Egyptian) faience, whichconsist of silica, alkali metal oxides and lime.Glass, glaze and enamel always contain largequantities of soda (Na2O) or another alkalimetal oxide, such as potash (K2O), andsometimes both, whereas Egyptian faiencecontains only quite small amounts of alkalimetal oxide. It has formerly been supposed,that because of the difficulty of reaching andmaintaining the high temperatures required tomelt glass from its raw ingredients, in ancienttimes, the raw ingredients were first formedinto an intermediate product known as frit.However, there is limited evidence for thispractice. In the fritting process, raw materials would be heated at temperatures just highenough to fuse them, and in doing so torelease carbon dioxide from the alkali carbon-ates. The resulting mass was then pounded topowder form (the frit). This was reheated athigher temperatures to form a semi-moltenpaste which could be formed into objects, or was heated at higher temperatures at which itcould melt to form true glass. A silicate glass is a material normally formedfrom silica, alkali metal oxides (commonly referred to as alkalis) and lime, when thesehave been heated to a temperature highenough to form them into a homogeneousstructure (formerly and ambiguously termed glass metal). Chemically, glass, glaze andenamel can all be identical in composition, thefundamental difference being their method of use in antiquity. The coefficient of thermalexpansion of a glass was not important whenit was used alone (unless it was applied on adifferent glass, as in the manufacture of cameoglass), whereas in a glaze or an enamel any difference in thermal expansion between themand the base on which they were fused couldcause the glaze or enamel to crack or becomedetached from the base material. In practice,glasses and enamels needed to have a lowmelting point, remain plastic as long as possi-ble while cooling and, apart from the very earliest glasses, be translucent or transparent(in contrast to the early glazing of earthenware where coloured decoration had been impor-tant). A glaze is a thin vitreous coating applied toanother material to make it impermeable, orto produce a shiny decorative appearance.Glaze was sometimes applied with the body material before firing, but more often it wasapplied to the object after it had received afirst firing, following which the object wasrefired to form the glazed surface.Faience is composed of fritted silica withabout 2 wt per cent of lime (CaO) and about 0.25 wt per cent soda, lightly held together with a bonding agent such as water. Theresulting paste was shaped by hand or in anopen mould and then heated until the limeand soda had reacted enough (fused suffi-ciently) to hold the silica particles together.During the formation process, faience objects formed a glazed surface with a similar compo-sition to the body, usually coloured blue orgreen with copper compounds. (Strictly speak-ing the term faience, derived from the nameof the Italian town of Faenza, should refer tothe tin-glazed earthenware made there.) Toreduce confusion the material discussed hereshould be referred to as Egyptian faience, orpreferably, glazed siliceous ware, (Nicholson, 1993; Smith,1996).The pigment known as Egyptian Blue, firstused in Egypt during the third millennium BC,and during the next 3000 years, in wall paint-ings, and as beads, scarabs, inlays andstatuettes, is the mineral (CaO.CuO.4SiO2) =(CaCuSi4O10). X-ray diffraction analysis hasshown that, in addition to this compound, theonly crystalline materials were quartz andtridymite (another of the crystalline forms of silica) (Chase, 1971; Titeet al., 1981). A enamel resembles a glaze in that it is alsofused to a body of a differentmaterial, in thiscase, metal; however, the term enamel is also usedto describe vitreous pigments used to decorateceramics and glass.
Chemical structure and composition
Zachariasen (1932) established that the atomsand ions in silicate glasses are linked togetherby strong forces, essentially the same as incrystals, but lacking the long range order which is characteristic of a crystal. Crystallinesilica (quartz) melts sharply at 1720°C from itssolid state, to a liquid, just as ice melts to form water at 0°C. This melting point is scientifically referred to as the liquidus. When the silicaliquid (molten glass) is cooled from above theliquidus, the randomly distributed molecules will endeavour to adopt a less random config-uration, more like those of crystals. However,an alternative three-dimensional structureforms because the crystallization process ishindered by the high viscosity of the glass,and the presence of the network modifiers.The melt becomes more and more viscous asthe temperature is lowered until, at about 1050°C it sets to form a solid glass (a stateformerly but no longer referred to as a super-cooled liquid). Moreover, the density of thatglass is less than that of the original quartz because there are now many spaces betweenthe ill-fitting molecules.However, in order to form a usable glass itis necessary to add certain oxides to the silica, which act as network modifiers, stabilizers andcolourants, and which also have a markedeffect on the structure of the resulting product. When network modifiers are added, they havethe effect of considerably lowering the viscos-ity of the melt. Thus there isthe potential for a different type of crystalcontaining atoms from the modifiers, to form inthe sub liquidus melt, provided the melt hasbeen held at the liquidus temperature for longenough. Thus a glass with the molar composi-tion 16Na2O, 10CaO, 74SiO2 can form crystalsof devitrite (Na2O.3CaO.6SiO2); which grow ata rate of 17μm per minute at a temperature of 995°C, the optimum temperature for growth of devitrite in that composition of glass. The totalchemical composition of the glass remainsunaltered (i.e. no atoms are added orsubtracted from those already in the glass),although the composition will change locally ascrystals of devitrite separate from the bulk glass. Ancient glasses have such complex compo-sitions that devitrification occurs much lesseasily than in modern glasses, so that if crystals of devitrite are present in a sampleundergoing examination, there may be doubtsconcerning the antiquity of the glass.However, the enormous block of glass madein a tank furnace in a cave at Bet She’arim, inIsrael, was found to be heavily devitrified(with the material wollastonite, CaSiO3) as aconsequence of containing 15.9 wt per cent of lime (Brill and Wosinski, 1965). The opalizingagent in some glasses may be a devitrificationproduct itself, which forms only when suitableheat treatment is given to the glass. Devitritedoes not occur as a mineral in nature.Early historians and archaeologists haveoccasionally used the term devitrification inquite a different sense, meaning loss of vitre-ous structure to describe glass that has weath-ered with loss of alkali metal ions, of otherconstituents of the glass and probably a gain in water content. This ambiguous use of the termshould be avoided (Newton and Werner, 1974).
Network formers
The principal network former in ancientglasses is silica (SiO2). Silicon and oxygen in crystalline silica (quartz) are arranged in adefinite pattern, the units of which arerepeated at regular intervals forming a three-dimensional network consisting of tetrahedra with a silicon atom at the centre and anoxygen atom at each corner; all four of theseoxygen atoms form bridges to silicon atoms of the four neighbouring silicon tetrahedra. Othernetwork formers are the oxides of boron(B2O3), lead (PbO) (Charleston, 1960) andphosphorus (P2O5). The presence of boron isimportant for clarifying glass compositions.However, it is difficult to analyse and so mighteasily be missed, especially since ancientglasses typically contained only 0.01 to 0.02per cent (whereas some Byzantine glassescontained 0.25 per cent boron). Boron enteredthe glass by way of the ash obtained by burning plants containing boric oxide. Themineral colemanite (hydrated calcium borate)(Ca3B6O11.5H2O) is found in western Turkey,and may have been used in glassmaking.The concept of network-forming oxides isillustrated in Figures 1.3 and1.4. Figure1.3 shows the regular structure of an imaginary two-dimensional crystalline material. Withinthe broken line there are 16 black dots (repre-senting atoms of type A) and 24 open circles(representing atoms of type O); hence theimaginary material has the composition A2O3 and its regular structure shows that it iscrystalline. If theimaginary crystalline material A2O3, shown in Figure 1.3, has been melted,and is cooled quickly from the molten state,the resultant solid might have the structureshown in
Figure 1.4. Here the broken lineencloses 24 black dots and 36 open circles andhence the composition is again A2O3 but thestructure is irregular and non-crystalline, repre-senting the amorphous, glassy or vitreous stateof the same compound. Note that theamorphous structure contains spaces and thusoccupies a greater volume than the crystallineone, and hence the crystal has a higher density than the glass, even though the chemicalcomposition is the same.
Figure 1.3, Schematic two-dimensional representationof the structure of an imaginary crystalline compound A2O3
Figure 1.4. Structure of the glassy form of thecompound in Figure 1.3.
Network modifiers
Figure 1.5 shows a structure which is nearerto that of silicate glass. It is again a simplifiedtwo-dimensional diagram, and the key to itnow mentions the word ion. Ions are atomsthat have been given an electrical charge, by adding or subtracting one or more electrons; cations having lost electrons, have a positivecharge, and anions having gained electrons,have a negative charge. The network-formingatoms are represented by black dots withinshaded triangles (atoms of silicon), and thenetwork modifying ions (positively chargedcations) are cross-hatched circles lying in thespaces of the network. Each network-formingtriangle (silicon atom) is accompanied by threeoxygen atoms (shown by small circles), whichcan be of two kinds. There are bridgingoxygen atoms (shown by plain open circles) which are shared between two triangles, thusjoining them together and forming part of thenetwork. There are also non-bridging oxygenions (shown by circles with a central dot) which belong to only one triangle; each of these thus bears a negative charge which isneutralized by a positive charge on one of thecross-hatched circles (cations). (Strictly, the Si-O-Si bonds are ‘iono-covalent’. They are notionic enough to refer to the oxygen as ions,and the Si as a cation. In the case of the Si-O non-bridging bonds, the Si-O bond is stilliono-covalent, but the negative charge on theoxygen gives it the ability to form an ionicbond to a cation in a nearby space.) It shouldbe noted that there is a very small amount of crystalline material in the diagram, near ‘A’ in Figure 1.5 , where four triangles are joinedtogether to form a regular (hence crystalline)area. (This can occur also in ancient glasses, where micro-crystallites can be detected.) Atall other points the triangles form irregularchains, which enclose relatively large spaces(and hence the density of the glass is less thanthat of a corresponding crystalline form).These spaces in the network have beencreated by the network-modifying cations which bear one or more positive electricalcharges, and which can be considered to beheld, by those electrical charges, to be more(or perhaps rather less) loosely bound in thoseenlarged spaces.The monovalent cations (which bear only one positive charge, having lost an electron toan adjacent non-bridging oxygen ion) areusually the alkali metal ions, either sodium(Na+) or potassium (K+), which bring withthem one extra oxygen ion when they areadded to the glass as soda or as potash.Because these cations bear only a singlepositive charge, they can move easily fromone space in the network to another (loosely bound). Thus, when the glass is placed in water, it becomes less durable because thecations (the smaller of the cross-hatchedcircles in Figure 1.5 ) can move right out of the glass into the water, thus making the waterslightly alkaline. In order to maintain theelectrical neutrality of the glass, these cationsmust be replaced by another cation such asthe oxonium ion (H3O).In the case of the divalent alkaline earth cations (the larger cross-hatched circles), eachbears a double positive charge (being associ-ated with two non-bridging oxygen ions, thecircles with dots inside). These are usually Ca++ or Mg++, added to the glass as lime(CaO) or as magnesia (MgO), but otherdivalent alkaline earth ions may also bepresent. The double electrical charge on themholds them nearer (more tightly bound) totheir accompanying non-bridging oxygens,making it much harder for them to move fromone space to another. Thus divalent alkalineearth cations play little or no part in carryingan electric current through the glass. Becausethey are associated with two non-bridgingoxygen ions, they strengthen the network,thus explaining why they help to offset thereduction in durability produced by the alkalimetal cations. However it should be noted thatin Figure 1.5 the double ionic linkages (tocircles with dots) are not immediately obvious.It is these linkages which determine the very different effects that the monovalent anddivalent cations have on the durability of glass.Notable advances have been made in theunderstanding of the structure of glasses. Forexample, it is now realized that the networkis actually loosened in the vicinity of themonovalent cations, channels (rather thanmerely larger spaces) being formed in whichthe cations can move even more easily than was formerly realized.
Figure 1.5
Schematic two-dimensional representationof glass, according to Zachariasen’s theory.
Phase separation
Despite the essentially homogeneous nature of bulk glasses, there may be minute areas, perhaps only 100nm (0.1m)in diameter, where the glass is not homogeneous becausephase separation has occurred. These regions(rather like that near ‘A’ in Figure 1.5 ) canhave a different chemical composition fromthe rest of the glass, i.e. the continuous phase(Goodman, 1987). Phase separation can occurin ancient glasses, and can have an effect ontheir durability, because the separated phasemay have either a greater or a lower resistanceto deterioration. The amount of phase separa-tion can be seen through an electron micro-scope.
Colourants
The coloured effects observed in ancient andhistoric glasses were produced in three ways:(i) by the presence of relatively small amounts(about one per cent) of the oxides of certaintransition metals, especially cobalt (Co),copper (Cu), iron (Fe), nickel (Ni), manganese(Mn), etc., which go into solution in thenetwork; (ii) by the development of colloidalsuspensions of metallic, or other insolubleparticles, such as those in silver stains (yellow)or in copper or gold ruby glasses (red ororange); (iii) by the inclusion of opalizingagents which produce opal and translucenteffects. The production of coloured glasses notonly depends on the metallic oxides presentin the batch, but also on the temperature andstate of oxidation or reduction in the furnace.Of course the exact compositions of ancientglasses were complex and unknown, beinggoverned by the raw materials and furnaceconditions, so that the results could not beacccurately determined.
Dissolved metal oxides/state of oxidation
Coloured glasses can be produced by metaloxides dissolving in the glass (similar to thecolours produced when the salts of thosemetals are dissolved in water), although theresultant colours will also be affected by the oxidizing or reducing
(redox) conditions inthe furnace. In the traditional sense, a metal was oxidized when it combined with oxygento form an oxide, and the oxide was reduced when the metal was reformed. The positioncan be more complicated when there is morethan one state of oxidation. For example, iron(Fe) becomes oxidized when ferrous oxide(FeO) is formed, and a blue colour isproduced in the glass (because Fe2+ ions arepresent), but it becomes further oxidized whenmore oxygen is added to form ferric oxide(Fe2O3), which imparts a pale brown or yellowcolour to the glass (due to the Fe3+
ionspresent). However, the situation is rarely sosimple and usually mixtures of the two oxidesof iron are present, producing glasses of various shades of green. When a chemicalanalysis of glass is undertaken, it is customary to quote the amount of iron oxide as Fe2O3,but that does not necessarily imply that all of the iron is in that state.The oxidation process occurs when an atomloses an electron, and conversely, reductiontakes place when an atom gains an electron.Consider the two reversible reactions set outin equations (1.1) and (1.2), where e– repre-sents an electron, with its negative charge. Inequation (1.1) the forward arrow shows thatan electron is lost when Fe2+ is converted to Fe3+.Fe2+ Fe3+ + e–
1.1 Mn3+ + e – Mn2+
1.2 The combined effects of equations (1.1) and(1.2) is equation (1.3), which shows that thereis an equilibrium between the two states of oxidation of the manganese and of the iron(Newton in Newton and Davison, 1989).Fe2+ + Mn3+Fe3+ + Mn2+
1.3But the Fe3+and Mn2+are the more stablestates, and hence the equilibrium tends to move to the right. Thus, when the conditions during melting of the glass are fully reducing(the equilibrium has been forced to the left,for example by producing smoky conditionsin the furnace atmosphere) the ironcontributes a bright blue colour due to the Fe2 ions (corresponding to FeO) and themanganese is in the colourless form so that ablue glass is obtained. When the conditionsare fully oxidizing (the equilibrium has beenmoved to the right by the addition of oxidiz-ing agents; by changing the furnace conditionsto have short, bright flames; or by prolongingthe melting time), the iron contributes abrownish yellow colour and the manganesecontributes a purple colour, so the glassappears brownish violet. When the conditionsare intermediate, a variety of colours areobtainable, such as green, yellow, pink, etc.including a colourless glass when the purplefrom the manganese just balances the yellowfrom the iron. This is the reason why, if thereis not too much manganese, it will act as adecolourizer for the glass which would other- wise be greenish in colour.These conditions have been experimentally studied by Sellner (1977) and Sellner et al.,(1979), who produced a forest-type glass in which the colouring agents were only manganese (1.7 wt per cent MnO) and iron(0.7 wt per cent Fe2O3). A variety of colours was obtained, from pale blue, when thefurnace atmosphere was fully reducing (withunburned fuel present and a very low partialpressure of oxygen in the waste gases)through green and yellow to dark violet whenthe furnace atmosphere was fully oxidizing(plenty of excess oxygen in the waste gases).Sellner et al . (1979) also examined samplesof glass excavated from two seventeenth-century glassworks sites, one at Glassborn/Spessart and the other at Hilsborn/Grünenplan, both in Germany. The composi-tions of the glasses at both sites were similarto each other, but the former factory hadproduced green glass and the latter hadproduced yellowish to purple glass. Measure-ments by electron spin resonance showed thatthe green glass had been melted under reduc-ing conditions and the Hilsborn glass hadbeen melted under oxidizing conditions. Thus,the colour of the glass had been determinedby its having been made using beechwood ash (which contains both iron and manganese),and the furnace atmosphere, and not by theaddition of manganese. The origin of colourin these glasses has also been investigated by Schofield et al. (1995), using synchrotronradiation.Greenish colours can be obtained fromcopper. For archaeological reasons it may benecessary to discover whether tin or zinc isalso present, because the presence of tin would suggest that bronze filings might havebeen added to the batch, whereas the presenceof zinc would suggest the use of brass waste.However, the presence of appreciableamounts of a particular oxide need not neces-sarily indicate a deliberate addition of thatmaterial. For example, Figure 1.6 showsremarkable differences in the potash andmagnesia contents of Egyptian Islamic glass weights, manufactured either before, or after,845 AD. Brill (1971a) suggested that the earlierexamples were made with soda from thenatron lakes, whereas the later ones couldhave contained potash derived from burntplant ash. There are still many problems andambiguities to be solved regarding the compo-sitions of ancient glasses, by analyses of samples from known provenances. However,there are many cases where the colouringagent is so strong that there is no problem. Figure 1.7 shows the contents of metal ions infive kinds of ancient glass; sometimes only
Figure 1.6
Chronological division of Egyptian Islamicglass weights into high- and low-magnesium types.(From Sayre, 1965).
(which contains both iron and manganese),and the furnace atmosphere, and not by theaddition of manganese. The origin of colourin these glasses has also been investigated by Schofield et al. (1995), using synchrotronradiation.Greenish colours can be obtained fromcopper. For archaeological reasons it may benecessary to discover whether tin or zinc isalso present, because the presence of tin would suggest that bronze filings might havebeen added to the batch, whereas the presenceof zinc would suggest the use of brass waste.However, the presence of appreciableamounts of a particular oxide need not neces-sarily indicate a deliberate addition of thatmaterial. For example, Figure 1.6 showsremarkable differences in the potash andmagnesia contents of Egyptian Islamic glass weights, manufactured either before, or after,845 AD. Brill (1971a) suggested that the earlierexamples were made with soda from thenatron lakes, whereas the later ones couldhave contained potash derived from burntplant ash. There are still many problems andambiguities to be solved regarding the compo-sitions of ancient glasses, by analyses of samples from known provenances. However,there are many cases where the colouringagent is so strong that there is no problem. Figure 1.7 shows the contents of metal ions infive kinds of ancient glass; sometimes only 0.02 per cent of cobalt is sufficient to producea good blue colour. The deliberate productionof an amber colour in ancient glass was in theform of iron -manganese amber describedabove, orcarbon-sulphur amber. (They can bedistinguished from each other because theFe/Mn colour has optical absorption bands at380 and 500nm, whereas the C/S colour hasits absorption bands at 430 and 1050nm.)The metals strontium (Sr), lithium (Li) andtitanium (Ti) enter glasses as trace elements inthe raw materials, in calcium carbonate forexample; beach sand containing shells is highin strontium in comparison with limestone which is low in its content, and therefore theamount present in glass is an indicator as to whether shell was a deliberate addition.Strontium is a reactive metal resemblingcalcium, lithium is an alkali metal resemblingsodium, but is less active and titanium resem-bles iron.
Figure 1.7
Colour element patterns in cobalt-blueglasses dating from the second millennium
BC. (FromSayre, 1965).
http://www.academia.edu/3386155/Conservation_and_Restoration_of_glass_Sandra_Davison