Archaeometric Analyses of European 18th–20th Century White Earthenware—A Review

Abstract

This study provides an overview of the few archaeometric analyses of European white earthenwares from England, France, Italy, Slovenia, and Switzerland. White earthenwares were an extremely successful mass-product between ca. 1750 and 1900. They became “the porcelain of the poor man” and replaced the older traditional pottery such as faïence. The invention of this new ceramic type took place simultaneously in England and France shortly before 1750. Contemporary recipes can be compared to the analytical results of these products. The ceramic bodies are, according to the chemical (X-ray fluorescence (XRF), scanning electron microscope-energy-dispersive X-ray spectroscopy (SEM-EDS), particle-induced X-ray emission-proton-induced X-ray emission (PIXE-PIGE) and microscopic analyses (scanning electron microscope-back scatter detector (SEM-BSE), artificial mixtures of clay + quartz + flux. Various techniques can be recognized. In England, a blend of a CaO-poor clay (with illite flux) + calcined flint is typical of the so-called creamware, whereas supplementary fluxes (Cornish stone) resulted in the Queen’s ware. In France and Central Europe, CaO-poor clays were mixed with grog (Paris) or with calcined flint/quartz pebbles/sand or with carbonates + Alk-/Pb-frits (Lorraine). Swiss CaO-rich bodies may contain bone ash or dolomite as flux. The products of the individual factories can be differentiated on the basis of their chemical composition. Micromorphological and chemical analyses revealed intensive physico-chemical reactions between the glaze and the body.

1. Introduction

White earthenware is a particular class of ceramics with a white, porous body and normally covered with a lead glaze. Such objects were manufactured for instance in France in the 16th century. The so-called Saint-Porchaire ware, a technologically outstanding Renaissance ceramic type, created during a brief period in the mid-sixteenth century [1], or the extra white pastes produced by Bernard Palissy (1510?–1590) in his Paris workshop in the years 1567–1586 [2,3]. However, these costly wares were restricted to the noble classes.

In the first decades of the 18th century, new and much cheaper bodies were invented roughly simultaneously in England and France [4]. English potters preferred CaO-poor clays, abundant in England, to create new ceramic types, but French potters used both CaO-poor and CaO-rich white firing raw materials. In both countries, production was soon running on an industrial scale, so the objects could be named white-bodied industrial earthenware.

In Staffordshire’s early 18th century thriving pottery district (Figure 1) local competition to whiten the pastes to emulate the whiteness of porcelain led around 1720 to the creation of the immensely successful white salt-glazed stoneware or common ware [5,6]. Between 1720 and 1740, further experiments with other raw materials such as the Al-rich and highly plastic Devon or Dorset ball clays gave birth to another body called creamware or cream colored ware, bisque fired at 1100–1200 °C [7,8]. The lead oxide used to glaze the body was contaminated with iron that imparted a creamy tint to the glaze thus giving it its descriptive name. Addition of calcined and milled flint improved the solidity and whiteness of the creamware. In his manufactory at Burslem, Josiah Wedgwood I (1730–1795) produced ca. 1761 a new kind of creamware body. He added kaolinite and feldspar to the paste and obtained a much whiter and harder body than the usual creamware. This was named Queen’s ware in 1765 [9], see Figure 2. Around 1775, cobalt oxide was added to the raw glaze mix to optically mask the yellow iron hue of the fired glaze. These objects were christened pearlware or china glaze ware.

French white earthenware originated during the years 1730–1750 in Central (Paris) und Eastern (Lorraine) France. On 27 June 1743, Claude-Imbert Gérin was granted a 10-year privilege to found a royal manufactory in Paris “de la fayance a l’imitation de celle d’angleterre” (of fayance imitating that of England) [4] (p. 84). It was established at the Rue de Charenton and transferred to a larger place near the bridge Pont-aux-Choux in 1749 [10,11,12]. This new paste was probably created as early as 1740. It was made of 66% Moret clay, a local white Al₂O₃-rich and CaO-poor clay from Montereau, and 33% calcined Moret clay (grog) [13] (p. 60). Its tableware was known for its typical rice-grain or barley-corn molded motif. This motif was probably derived from English salt-glazed wares, copied 1750–1756 at Pont-aux-Choux (Figure 3a) and in use until the closing of the factory in 1788 [4,14]. In Lorraine, Jacques II Chambrette managed to produce in the town of Lunéville a CaO-rich (calcareous) white earthenware, called terre de pipe (Figure 3b), perhaps as early as 1731 [15]. His success triggered the creation of 23 manufactories in Lorraine during the 18th century [4,15]. Their whitewares were generally made up of four main ingredients: (1) Al₂O₃-rich refractory clay, (2) chalk, (3) Pb-frit, and (4) calcined flint or quartz pebbles [16]. Three periods can be distinguished in the historical evolution of French white earthenware [4]: (1) 1743–1790: the coexistence of Al₂O₃-rich & CaO-poor and CaO-rich white earthenware bodies, (2) 1790–1830: the improvement of the non-calcareous wares by adding calcined flint or quartz pebbles to the paste, resulting in the so-called cailloutage body, and (3) after 1830: a further improvement of the non-calcareous bodies by admixing kaolinite and feldspar to the paste [4]. Visibly, the French potters could not compete as they were technologically at least 50 years behind. In the late 18th and early 19th centuries, many white earthenware factories blossomed in France and the neighboring countries such as Germany [17], Italy [18], and Switzerland [19].

2. Nomenclature

The English (stonepaste, creamware, Queen’s ware), German (Steingut) and Italian (terraglia) nomenclature of this particular class of ceramics, made up of synthetic and porous pastes turning white during firing, is rather simple. In France, however, the nomenclature is more complex.

2.1. French Nomenclature in the 18th and 19th Centuries

The oldest book dealing with this kind of material was published anonymously [20], but attributed by Brongniart to Oppenheim [21]. He quotes that the white pottery in the English manner is a faïence fine covered with a transparent glaze, known as fine pottery, white faïence, English earthenware, terre de pipe, English faïence, white earthenware, and cailloutage. The term terre de pipe is a synonym for English faïence and was continuously used in the French technical treatises of the early 19th century [20,22,23]. Brongniart (1770–1847), director of the Royal porcelain manufactory of Sèvres from 1800 to 1847, classified white earthenwares into three types: (1) the so-called terre de pipe (“faïence fine marnée”) with a calcareous body; (2) the creamware or cream-colored earthenware (“faïence fine cailloutée, cailloutage, terre anglaise, poterie façon anglaise”), with a CaO-poor (non-calcareous) body; and (3) the semi-porcelain (“porcelaine opaque, ironstone, faïence fine dure, faïence fine feldspathique, lithocéram”), also with a CaO-poor body [23]. Evidently, the French terminology is complex and confusing [4,16,24,25]. The actual most popular French term faïence fine—designating a white, porous body covered with a transparent lead-glaze—should be replaced by the more neutral term terre blanche, since a French faïence has a colored, porous body and a white tin (cassiterite, SnO₂) opacified glaze.

2.2. Whitewares, White Earthenwares and Porcelains

Clay-derived ceramic objects with a white body (whitewares) belong either to the family of white earthenware, if their water adsorption is above 2%, or to the porcelains, with water adsorption below 2% [26].

2.3. Archaeometric Nomenclature

A priori, scientists know nothing about the ingredients of a whiteware’s paste. The nomenclature should therefore be based on the microscopic aspects and the chemical analyses of the final product. Irrespective of the specific type, whitewares are made up of two or three fundamental ingredients: (1) plastic raw materials such as kaolin, China clay or ball clays (highly plastic, refractory and kaolinitic clays); (2) non plastic materials such as calcined and finely milled flint, quartz pebbles and quartz-rich sands or ground fired clay (grog); and (3) fluxing materials such as illite, lime, chalk, limestone, dolomite, magnesite (giobertite), feldspar, Cornish stone, pegmatite, lead frit, alkaline glass, fusible sand and bone ash. In practice, one or more of these three ingredients can be mixed.

All chemical analyses revealed that a white earthenware body consists of three to four dominant oxides: SiO₂, Al₂O₃, CaO, and MgO. The classification can therefore be based on these four constituents. A white earthenware with 80–85 wt % SiO₂ or higher will be classified as siliceous white earthenware. Its body consists mainly of crushed quartz grains with little or no clay, all welded together by a vitreous flux [27,28,29]. This pottery type is also named stonepaste.

A white earthenware with SiO₂ lower than 80–85 wt % belongs either to the aluminous white earthenwares (Al white earthenware), with no or very little CaO and MgO (maximum 3–4 wt %), or to the CaMg white earthenwares with conspicuous amounts of CaO and MgO. Both oxides are derived either from carbonate rocks (limestone, dolostone) or from carbonate minerals (calcite CaCO₃, dolomite CaMg(CO₃)₂, Magnesite MgCO₃) or their mixed end members. The nature of the carbonate ingredients can be ascertained with a MgO-CaO plot (Figure 4).

Pastes with magnesite alone should plot on the ordinate, but in reality, there is always some CaO present that will enlarge the ordinate to the red area A. A high dosage of magnesite results in a placement of the analysis towards the very top of the ordinate or the red area, a weak one by a downward shift on that line/area. The central black line C connects the zero point with the projection pole of the mineral dolomite, which lies here outside the diagram. Ceramic bodies plotting on this line show calcium and magnesium oxide content, which are explained by the addition of the mineral dolomite or the rock dolostone to their paste exclusively, without other ingredients containing calcium or magnesium in their chemical formula. On the other hand, if the calcium is only due to the presence of the mineral calcite (or of a rock containing this mineral like chalk, limestone, marble) or of another Ca-phase like gypsum CaSO₄·2H₂O, in the recipe, without or very little MgO, all the points will be on the horizontal abscissa or in the blue area E. The compositions dependent on the simultaneous presence of these carbonates in the paste’s recipe will have their projection point in the areas between the black “dolomite” line and the abscissa or the ordinate. The space B to the left of the black central line will accommodate all compositions derived from mixtures of Ca- & CaMg- & dominant Mg-carbonates, the space D mixtures with dominant Ca-carbonates. This binary diagram allows the classification of the CaMg white earthenwares in four groups: (A) magnesitic, (B) MgCa; (C) dolomitic; (D) CaMg, and (E) calcitic white earthenwares.

3. Production of White Earthenwares: Historical and Technical Aspects

3.1. Some Manuscript Sources

No such sources seem to have been published for England. In France, the oldest recipes can be found in the short chapter “De la terre dangleterre blanche, et marbrée” in Pierre Paul Caussy’s manuscript, probably written in the time span 1735–1747 [30]. He coined the term terre de pipe. Terre de pipe recipes of other French manufactories (Aprey, Audun-la-Tiche, Jussy Saint-Clément and Septfontaines) are known [4,24,31]. A summary is given in [16] (p. 768). Some Swiss recipes of the 18th to the 19th century are reported for CaO-rich white earthenwares produced in Carouge [32], Lenzburg [33], Les Pâquis [34], Luzern [35], Matzendorf [35], and Nyon [32]. Late 19th century recipes for CaO-poor white earthenware of the Carouge manufactory are presented in [36,37]. As for Italy, the recipe of a CaO-rich white earthenware is known for the Torino manufactory [37,38].

3.2. The Production Technique of White Earthenware According to Oppenheim [20]

In the title of his book, Oppenheim calls himself a “former manufactory director”, but we know nothing about his life. He gives a detailed account of how to produce poterie blanche, façon anglaise; i.e., CaO-poor white earthenware.

According to him, the mixing of clay and flint is the most sensitive phase of the preparation. Not enough flint would cause a high quantity of pieces to break after firing, and too much of it would produce fragile pieces due to their high porosity. For France, the name of about fifteen places where white firing clays can be found are given, along with places near Antwerp and in the region of Cologne (Germany). The presence of chalk in the clay could cause a distortion of the shape during firing. With Montereau clay, first fired at a temperature of 100° Wedgwood, one sevenths part of flint is added, according to the following recipe (in parts): 87 of washed plastic Montereau clay + 13 of pounded flint [23]. After drying, the unfired pieces are placed in saggars made of clay containing neither iron nor chalk, including a strong proportion of grog of a similar composition.

The first (bisque) firing occurs in a cylindrically formed kiln with a dome roof. It has four to eight grated fireboxes placed outside. In France, the fuel is wood, in England coal. After the low fire (petit feu), which lasts two to three hours, more wood of a larger size is added for another thirty to forty hours, then comes the hard firing (grand feu), a period of twenty to thirty hours characterized by a long and intense flame obtained by laying the wood in embankments. The inside of the kiln has then turned white and the firing of the ware is over. In summary, the first firing lasts fifty to seventy hours in all—the more flint there is, the longer the firing. When firing is terminated, the kiln is left to cool down.

The glaze is a mixture of ground calcined flint, minium (Pb₃O₄) and potash, an impure K-carbonate obtained through the combustion of wood. For a Montereau clay with 1/7 calcined flint, first fired at a temperature of 100° Wedgwood, the glaze recipe is (in parts): 8 of white sand + 10 of minium + 5 of potash and 1/1000 of the whole of cobalt. The mixture is placed in a closed crucible and fritted in the kiln. Another recipe consists in making a preliminary frit with white sand and an equivalent quantity of minium mixed with two parts of commercial potash, from which two glazes can be obtained. Glaze 1 (parts): 2 of vitrified mass + 1.5 of minium + 1/16 of lead-free white glass, all fired at 12° Wedgwood. Glaze 2 (parts): 2 of vitrified mass + 1 of lead oxide (litharge d’or) + 1/16 of white glass, all fired at 20° Wedgwood.

The second (glaze) firing occurs in a kiln similar to the first one, but smaller. The firing time only lasts from sixteen to twenty-four hours, with twelve to eighteen hours of low fire and four to six hours of hard firing periods. The firing temperatures are lower than those of the first firing, and do not surpass 15–18° Wedgwood for the French glazes and 27–30° Wedgwood for the English ones [22].

3.3. French and German Technical Handbooks of the First Half of the 19th Century

Other French technical commentaries followed in 1827 [23] and 1830 [22]. They do not differ from Oppenheim, giving only some minor details about the French technique (cailloutage) for the English creamware type. Bastenaire Daudenart’s book was soon translated into German by the director of the Royal porcelain manufactory of Berlin [39]. Boudon de Saint-Amans, who worked in the Royal porcelain manufactory of Sèvres, gives a thorough explanation of how creamware was produced in England [40]. In 1835, another treatise in German was published [41]. Brongniart’s first edition of his masterpiece work appeared in 1844 finally [21]. It was to become the reference book for French potters. For France, he focused mainly on CaO-poor white earthenware with some mention to Lunéville, Mettlach, Sarreguemines, Saint-Clément and Vaudrevange as places where CaO-rich (calcareous) earthenware was still produced. Only one recipe—Brongniart does not say where it comes from—is given for such a paste (in parts): 85.4 of clay (75 wt % SiO₂, 25 wt % Al₂O₃) + 13 of flint + 1.6 of lime. Visibly, this is a creamware and not a calcareous paste as it contains very little calcium. He compiled eight recipes of English white earthenwares (creamware, Queen’s ware) consisting of differing mixtures of ball clays, clays, kaolin, calcined flint, altered feldspar, altered pegmatite and shavings, summarized in [25] (p. 158). In the first decades of the 19th century, CaO-rich white earthenware bodies were obviously antiquated and not worth reporting in technical treatises.

3.4. Firing Temperatures

The temperatures of the first and second firing were estimated in the 18th and 19th centuries with the so-called Wedgwood pyrometer [23,42]. In

Table 1. Firing temperatures.

Type	Flux	CaO & MgO	Bisque Firing		Glaze Firing
		wt %	Pyrometer	oC	Pyrometer	°C
CaO-poor	Clay		[23]: 90–100	1200–1250	[23]: ca. 12, 27–30	900
			[23,42]: ca. 60		[42]: 20–25
	Feldspar		[42]: 89–100	1140–1300	[23]: 27–30	1000–1140
CaO- & MgO-rich	Ca- and Mg-carbonate	4–7		1060–1050		900–1000
		13–14	[23]: 25–30, [42]: <25–30	ca. 980	[42]: <25–30	ca. 980

, these assumptions are compared with the °C of today [43].

4. Analytical Studies—A Summary

A discussion of the chemical analyses of Saint-Porchaire and Palissy ware bodies and glazes can be found in [6]. This chapter will focus on all published scientific analyses of English, French, Italian, Slovenian, and Swiss white earthenwares (mostly tableware).

4.1. English White Earthenware

There are surprisingly few archaeometric studies of English specimens. The chemical composition of a Staffordshire (Wedgwood?) paste was published as early as 1830 [22] and a single Wedgwood pottery analysis was reported in 1857 [42]. Recent studies dealt with five sherds of unmarked eighteenth-century Staffordshire white bodies of unspecified typology and provenance (creamware, Queen’s ware, pearlware, consumption site, production site, collection?) [44], and three fragments bearing the Wedgwood impressed stamp from archaeological findings in Bern, the Swiss capital, and produced between 1769 and 1832 [25].

4.1.1. Body: Microstructures, Chemical Compositions and Technical Aspects

The 18th century Staffordshire bodies show “abundant angular quartz in fine-textured, interconnecting relict-clay-glass phases” [44]. Scanning electron microscope-energy-dispersive X-ray spectroscopy (SEM-EDS) analyses reveal (wt %) 76.2–78.6 SiO₂, 17.2–19.6 Al₂O₃ and 1.5–1.8 K₂O, the other oxides having maximum values not higher than 1 (

Table 2. Chemical compositions (Min.–Max.) of English and French CaO-poor white earthenwares. Oxides and chlorine in wt %, other elements in ppm.

	Staffordshire	Wedgwood		Pont-aux-Choux	Creil	Sarreguemines
Date	18th century	1769–1832	?	End 18th century	1797–1832	ca. 1820
N analyses	5	3	1	5	2	1
Method	SEM-EDS	XRF	PIXE-PIGE	XRF	XRF	XRF
Reference	[44]	[25]	[45]	[16]	[25]	[16]
SiO2	76.2–78.6	73.7–75.52	70.2	60.72–63.29	77.02–79.37	61.14
TiO2	0.4–1.0	0.75–1.08	0.65	0.74–0.92	0.59–0.62	1.06
Al2O3	17.2–19.6	18.45–21.74	20.6	31.49–34.02	17.02–19.18	29.95
FeO, Fe2O3	0.8–1.0	0.54–0.68	0.5	1.17–1.29	0.65–0.71	1.43
MgO	<0.2–0.4	0.11–0.14	0.35	0.01–0.12	0.08–0.10	0.52
CaO	0.4–0.5	0.50–0.90	3.50	0.57–0.75	0.4	0.38
Na2O	<0.2–0.5	0.26–0.33	0.42	0.83–0.94	0.19–0.29	1.95
K2O	1.5–1.8	0.92–1.06	1.35	0.61–0.76	0.34–0.37	3.60
P2O5		0.05–0.06		0.35–0.71	0.12–0.13	0.10
SO3			1.62
Cl			0.23
Ba		203–226		261–354	128–132	646
Cr		76–85		157–208	87–90	246
Cu		27–67		2–12	10	33
Nb		12–20		17–23	17	20
Ni		17–25		76–104	18–19	66
Pb		98–643	4641	654–2041	1744–3764	503
Rb		49–67		34–52	11–17	229
Sr		89–106		642–1156	211–227	148
Y		19–22		49–67	53–73	21
Zr		159–193		212–228	101–110	191
Zn			100

).

The microstructure (Figure 5a) of the three Wedgwood specimens (consumption site, marked) consist of: (1) angular SiO₂ fragments with maximum diameters of 40–50 µm; (2) Rare inclusions type A, B, and C; and (3) a fine-grained, porous, and felt-like matrix (Figure 5). The SiO₂ particles are characterized by a multitude of internal cracks and sub-domains in the range of ca. 2 µm. They consist of quartz, as shown by the X-ray diffraction (XRD) recordings. Other SiO₂ polymorphs such as cristobalite or tridymite may well be present, but their amounts would then be below the XRD detection limit. The peculiar internal structure is characteristic of calcined flint, because calcined quartz pebbles or quartz sands show a homogeneous microstructure (see below). Inclusion type A and B have typical features of grog or ceramoclast [46], such as different chemical composition and microstructure compared to the surrounding body, angular shapes, and drying cracks around the inclusion [47,48]. The differences in their chemical composition as compared to the Wedgwood bodies exclude that they may represent former bisques or paste shavings. The angular type A fragments show a glassy matrix with embedded needles of mullite, similar to highly fired hard paste porcelain bodies. Their chemical composition is compatible with a derivation from former Si-rich ball clays or a mix of those clays with Cornish stone. The type B inclusions are rounded to subangular with a phyllosilicate textural aspect. These microstructural aspects and their higher alumina content compared to the type A inclusions indicate their derivation from a kaolinite rich (high Al) clay, that is an Al₂O₃-rich ball clay. Both types were fired and crushed before admixing to the matrix clays. The type C inclusions are round pores of former carbonate inclusions (limestones) with characteristic calcium-rich rims up to 12.7 wt % CaO, relicts of the original clay(s). The matrices, that is, the former clays, of the Wedgwood samples are chemically similar and are made up of dominantly two oxides, SiO₂ and Al₂O₃. Their overall SEM-EDS analytical data matches those of kaolinitic-illitic clays of the ball clay type. They may represent: (1) a single, SiO₂-rich ball clay; (2) or a batch of one or more ball clays; (3) or a batch of one or more ball clays with one or more china clays. The TiO₂-values suggest some china clay admixture. However, this is not supported by the Al₂O₃ content, which is too low for such an ingredient.

The X-ray fluorescence (XRF) bulk body compositions of the Wedgwood samples show comparable silica (73.70–77.29 wt % SiO₂) and alumina (18.45–21.74 wt % Al₂O₃), but lower potassia (0.92–1.06 wt % K₂O) compared to those of the Staffordshire bodies (SEM-EDS analyses). As outlined before, Josiah Wedgwood I creamware is a mix of two ingredients (ball clay + calcined, ground flint), and his Queen’s ware a blend of calcined flint + ball clay + china clay + Cornish stone. China clays have significantly more alumina, and Cornish stones more Alkalis, than Ball clays [25]. This raises the following question: do the analyzed Wedgwood objects belong to the creamware or the Queen’s ware category? If the analyzed pieces pertained to the latter, they should have higher Al₂O₃, Na₂O and K₂O values than the measured ones, see the discussion of true Queen’s ware (faïence feldspathique, [43]) in Section 4.2.1. For this reason and because their chemical composition is similar to that of a Wedgwood cream colored earthenware from 1759, with, in wt %, 76.10 SiO₂, 20.45 Al₂O₃, 1.00 Fe₂O₃, 0.75 CaO, 0.14 MgO and 1.60 Alk [42], one may favor a creamware classification. On the other hand, the microstructural analysis (inclusions type A and B) speaks against this. These objects could therefore document the beginning of the transition from the creamware to the Queen’s ware type, with little kaolinite and Cornish stone admixtures.

A particle-induced X-ray emission-particle-induced gamma-ray emission (PIXE-PIGE) analysis revealed deviations from the XRF results [45]. CaO is significantly higher (3.5 wt %) and SO₃ and Cl were detected in some amounts (Table 2). One might therefore suspect an admixture of a calcium-containing material (gypsum according to the sulphur?) with some rock salt, but such ingredients were not found in the eight English recipes mentioned by Brongniart and reported in [25] (and Table 6 therein).

The high concentrations of lead (max. 643 ppm Pb) in the Wedgwood bodies contrast with the very low level of this element in the raw materials (Table 2, XRF). These do not exceed 10 ppm in carbonates (limestones, dolomites) or 110–240 ppm in clays [49,50]. Such high lead contents may be due to: (1) an addition of Pb glass (frit) during the preparation of the paste; (2) an infiltration of the aqueous suspension containing the crushed glaze during their application on the porous bisque, between the first and the second firing; (3) an infiltration of the molten lead glaze into the bisque during the second firing; (4) an infiltration of Pb vapors into the ceramic objects during the firings or (5) an insufficient abrasion of the glaze during sample preparation. The fifth hypothesis is unlikely, because great care was taken to completely remove the vitreous film. On the other hand, no fragments of a Pb frit or a Pb bearing grog could be observed in the bodies. Therefore, Hypothesis (1) can be dismissed too. No argument can be put forward to decide between the remaining hypotheses. The fact that Pb infiltration has effectively taken place is shown by the profile measurements (see below), but it cannot be decided to what extent the three remaining mechanisms can each be held responsible.

The PIXE-PIGE analysis revealed surprisingly high Pb concentrations (Table 2). These are usually related to the presence of a Pb frit in the body (see below: French CaO-rich bodies of Lorraine). It would be very surprising that Wedgwood mixed lead frits into his paste. It is therefore more likely that this non-destructive method simultaneously analyzed body together with some lead glaze.

The XRD phase association is characterized by the high temperature phases cristobalite and mullite. Experimental firings of refractory, Ca- and Fe-poor clays reported mullite formation at temperatures above 950 °C and cristobalite crystallization at temperatures starting at 1050 °C [51,52]. Such high temperatures are in good agreement with the upper limit of ca. 1200–1250 °C for the first firing stage of creamware or cailloutage bodies [43].

4.1.2. Glazes: Microstructures, Chemical Compositions and Technical Aspects

The transparent glazes used for the Staffordshire white earthenware are of the high lead type (48.3–57.8 wt % PbO) containing a considerable amount of silica (35–43.9 wt % SiO₂), small percent of alumina (3.9–6.1 wt % Al₂O₃) and negligible alkali (K₂O + Na₂O below 1 wt %).

The Wedgwood glazes are typically 50–80 µm thick and appear homogenous. There are rare gas bubbles and crystalline phases of (1) subangular and rounded SiO₂ grains, interpreted as relict quartz; (2) tiny rectangular SiO₂ crystals, often in clusters or surrounding quartz cores, interpreted to be cristobalite because of their crystal habitus. An irregular and undulated intermediate zone with a thickness of ca. 5–20 µm, consisting of small (less than 2 µm) euhedral, rectangular and zoned crystals rich in alumina, silica, and lead can be observed in the glaze close to the sharp contact with the body. The glaze main components are PbO, SiO₂ and Al₂O₃ with minor amounts of K₂O, Na₂O, MgO, CaO, TiO₂, and Fe₂O₃ (

Table 3. Thicknesses and scanning electron microscope-energy-dispersive X-ray spectroscopy (SEM-EDS) analyses (wt %) of transparent glazes of English and French CaO poor white earthenwares.

	Staffordshire	Wedgwood	Pont-aux-Choux	Creil	Sarreguemines
Date	18th century	1769–1832	End 18th century	1797–1832	ca. 1820
N samples	5	3	2	1	1
Reference	[44]	[25]	[16]	[25]	[16]
Thickness (µm)		50–80	2–260	5–150	25
Na2O	<0.2–0.7	0.1–0.7	1.5–2.4
MgO	<0.2–0.4	0.2–1.2	0.1–0.3		0.2
Al2O3	3.9–6.1	4.5–7.4	3.4–3.5		7.8
SiO2	35.0–43.9	39.2–41.7	32.6–39.1	25.9	30.5
Cl				0.5
K2O	<0.2–0.4	0.0–0.8	1.1–2.2	0.3	0.6
CaO	0.3–0.8	0.1–0.7	1.4–2.3	0.4	0.1
FeO, Fe2O3	<0.2–0.8	0.3–0.5	0.5–0.6	0.3	0.4
TiO2	<0.2–0.4	0.2	0.1		0.2
PbO	48.3–57.8	48.6–52.6	50.2–57.3	72.6	60.2

). Lead infiltration was detected from the glaze/body interface inwards, affecting the outmost 80 µm of the bodies (5.9 wt % PbO in the body close to the contact).

Glazes and ceramic bodies interact during the last firing with complex digestion/diffusion processes [53]. Such processes must have affected the studied samples, as revealed by the important lead diffusion from the glaze into the body. Alongside with this glaze-body contamination, the reverse occurs; i.e., elements diffuse from the ceramic body towards the glaze. Alumina, for instance, migrates in significant concentrations (over 1 wt %), as shown by the experimental work of [54], into an alumina-free lead glaze applied on kaolinitic bodies as far as 150–200 µm. For the studied samples, such an Al-contamination could have theoretically affected the whole glaze section considering their thicknesses (50–80 µm). However, the Al-profile from the contact body/glaze into the glaze revealed a body derived alumina contamination only over a distance of 35 µm. The high Al₂O₃ values measured in the uncontaminated outermost parts must result from the addition of clay or feldspar to the glaze mixture. Unfortunately, the literature does not mention the recipes for Wegdwood’s white earthenware glazes.

4.2. French White Earthenware

As for England, the archaeometric studies of 18th and early 19th French white earthenware are sparse. There is one old Lunéville terre de pipe analysis [42] and another of a bisque from the manufactory Bosch at Luxemburg (1824) [43]. Their iron oxide contents (1.2–2 wt %) are obviously very high for a white earthenware, which is probably a method-relevant artifact. More recently, seven specimens from two manufactories in and outside Paris (Pont-aux-Choux, n = 5; Creil, n = 2) as well as 19 objects attributed to six, mostly Lorraine manufactories (Bois d’Épense, Lunéville, Niderviller, Saint-Clément, Sarreguemines and of unknown provenance) were studied [16,25]. Examples of French CaO-poor white earthenware are given in Figure 6. The origin of the five Pont-aux-Choux pieces was left open [16]. Meantime, Christian Maire considered the linear detailing between the grains in the rice pattern of the five studied objects [55] to be characteristic of this manufactory. The mean values of two groups (production date 1765–1780, ca. 1820) from the manufactory of Meillonnas can be found in [56] and one single analysis from the small workshop Jussy on the southern shores of lake Geneva in [37].

4.2.1. Body: Microstructures, Chemical Compositions and Technical Aspects

The five unmarked samples (from a museum collection destroyed by a fire) with a rice-grain molded decoration typical for the Pont-aux-Choux factory (Figure 6b) exhibit a grog rich body (Figure 5b). The quartzes were already present in the clay, as exemplified by their rounded shapes. These samples differ from the Wedgwood objects by a very high alumina content (Figure 7a). Their microstructure is in good agreement with the paste recipe of Pont-aux-Choux; i.e., a mix of local Moret clay with its calcined counterparts in a ratio 66:33, as reported by Hellot (quoted by [13]) (p. 60). Moret clay is a variant of the widespread Montereau clay [25] and has been characterized in 1778 as “… little sandy, very binding & very refractory. The best and purest of all …” is from a great dig on the way to the mountain of Moret. A pretty pottery of white earthenware of the English type is made in Montereau, and even more in Paris, at the manufactory of Pont-au-choux” [57]. Grog and bodies are as expected CaO-poor (<1 wt %), but richer in Al₂O₃ as the published chemical analyses of Montereau clays (max. 31 wt % Al₂O₃ [25]), see [16]. The chemical composition is clearly different from that of the Wedgwood tableware (Table 2). The high alumina and low silica match the microscopic appearance. However, the Sr values are surprisingly high and not yet explained.

Both Creil specimens (consumption site, marked pieces) differ chemically (Table 2) and microstructurally from the Pont-aux-Choux products. They are like those of the Wedgwood ware, but without grog inclusions (Figure 5c). This fits in well with this kind of body, named cailloutage, where calcined and crushed flint (“cailloux”) were admixed to a non calcareous Montereau clay in the typical English creamware tradition [20,23]. Their elemental composition, however, is very distinct from the three Wedgwood samples, as they show considerably lower amounts of TiO₂, K₂O, Ba, Cu, Rb, Zn, and Zr, and higher contents of Sr and Y (Table 2). With the help of the Y-Zr binary diagram, they can well be distinguished from the Wedgwood ware (Figure 7b).

The stamped bowl from Sarreguemines (private collection) has a rare pink luster Burgos, characteristic of this manufactory in the 1820’s. Its microstructure differs significantly from that of the objects from Creil and the Wedgwood manufactories and is similar to that of Pont-aux-Choux, although no grog could be observed (Figure 5d). Their chemical composition differs from that of Pont-aux-Choux (Table 2), and both products can easily be differentiated as shown in Figure 7. Na₂O and K₂O are much higher in the Sarreguemines sample compared to the Pont-aux-Choux specimens. The raw clays were therefore different in both factories.

The CaO-rich unmarked bodies (museum and private collections,

Table 4. Chemical compositions (Min.–Max.) of French CaO-rich white earthenware. Oxides in wt %, elements in ppm.

	Bois d‘Epense	Lunéville	Meillonnas		Niderviller	Saint-Clément	Sarreguemines	Jussy
Date	End 18th century	ca. 1780–end 18th century	1765–1780	ca. 1820	ca. 1770, 1793–1802	1755–1760, ca. 1770, end 18th century	ca. 1805	1824–1839
N analyses	1	5	9	2	2	9	1	1
Method	XRF	XRF	XRF	XRF	XRF	XRF	SEM-EDS	XRF
Reference	[16]	[16]	[55]	[55]	[16,25]	[16]	[16]	[37]
SiO2	66.95	68.47–70.07	67.45–74.86	60.17–61.19	64.59–66.87	57.15–63.72	73.0	57.06
TiO2	0.61	0.40–0.54	0.48–0.57	0.56–0.57	0.45–0.65	0.60–0.83	0.7	0.76
Al2O3	14.39	9.14–10.99	10.53–13.35	15.00–15.55	12.94–16.44	15.39–19.33	12.2	14.34
Fe2O3	0.82	0.45–0.56	1.08–1.29	1.36–1.62	0.90–0.93	0.60–0.82	0.7	0.77
MnO	0.02	0.01–0.02	0.01–0.02	0.01–0.04	0.01–0.02	0.01–0.02
MgO	0.36	0.55–0.81	0.63–0.84	6.23–7.61	0.57–0.60	0.36–0.46	0.5	9.18
CaO	13.64	12.12–14.56	7.43–14.95	12.90–13.65	10.49–16.81	14.56–17.03	10.7	13.31
Na2O	1.39	2.72–2.32	0.42–1.70	0.06–0.10	1.06–2.13	0.34–1.39	0.8	0.39
K2O	0.61	1.26–2.01	1.21–2.03	1.28–1.49	0.91–1.58	1.69–1.95	1.4	1.89
P2O5	0.10	0.11–0.26	0.14–0.22	0.16–0.26	0.14–0.15	0.05–0.10		0.04
Ba	108	200–255	153–189	189–259	302–420	226–432		277
Cr	61	57–91	14–62	70–71	69–123	88–123		84
Cu	46	17–71			27–32	16–116		28
Nb	14	11–12			11–12	16–19		16
Ni	17	17–26	15–25	24–31	26–47	29–117		35
Pb	11,840	2525–12,964			3657–7424	11,681–27,880		12,736
Rb	9	44–120	53–119	72–77	58–74	11–91		71
Sr	380	246–327	220–367	94–118	294–389	262–341		86
Y	45	16–26			18–36	34–57		42
Zn	90	5–14	1–21	10–25	27–29			61
Zr	182	128–165	263–319	257–265	126–162	183–256		185

) from Lorraine and the neighboring Bois d’Épense manufactory are composed of (Figure 8): (1) angular SiO₂ fragments (=quartz) with diameters not exceeding 340 µm. As no internal cracks and sub-domains typical for calcined flint grains can be observed, it is assumed that they most probably correspond to calcined sand or quartz pebbles. This contradicts the currently known recipes, which list only flint, but no quartz pebbles or quartz sand [16]; (2) roundish to angular lead-silica frits; (3) CaO-rich particles and more or less circular pores corresponding to primary particles rich in CaO—that is, carbonate or portlandite Ca(OH)₂—having reacted with the clay paste during the firing; and (4) a fine-grained matrix [16,25]. Angular potassium feldspar and grog grains can also be found in some bodies.

Three frit types can be defined according to their microstructure and chemical composition: (1) cassiterite-bearing PbO-SiO₂ frit (rare), Figure 8a; (2) low-Ca PbO-SiO₂ frit (Figure 8b and Figure 9); and (3) high-Ca PbO-SiO₂ frit. Two SiO₂-polymorphs can be observed in the cassiterite-free frits: rounded individual grains or cores of SiO₂-clusters, interpreted to be relic quartzes, and idiomorphic laths, probably cristobalite in view of the crystal habitus, either as clustered individuals or surrounding quartz cores. The relative amounts of glassy and crystalline phase(s) in the frit particles are highly variable. The contact with the surrounding matrix is well defined and sharp. With few exceptions, no Pb diffusion from the frits into the surrounding ceramic matrix could be measure. The overall chemical composition of the matrix; i.e., the former clay—is close to that of kaolinitic-illitic clays. According to modal analyses, the amount of former clay (matrix) in these artificial bodies shows less variation (57–70 vol %) than the quartz (6–27 vol %), the CaO-rich area (0.6–8.5 vol %) and the frit contents (3–8 vol %).

Stylistic and Cluster analyses attribute 17 tableware samples to five manufactories (Bois d’Épense, Lunéville, Niderviller, Saint-Clément, Sarreguemines) while two samples could not be assigned to any manufactory. Considering the products of Sarreguemines, it reveals a change in the recipe between 1805 and ca. 1820, with a transition from CaO-rich to a CaO-poor body (Table 3 and Table 4). Further analyses of well-dated objects could define this change even more precisely.

The faïence manufactory Meillonnas also produced CaO-rich dishes in both 18th and 19th centuries (Table 4). The studied samples are from excavations on the production site and include unmarked bisque and glazed fragments [56]. The largest chemical difference is found in the MgO concentration, which is below 1 wt % in the older group, dated 1765–1780, but reaches 6–7 wt % in the younger group (Early 19th century). From the 18th to the 19th century, therefore, a significant change in the recipe must have taken place. Although there are no SEM analyses of these objects, it can be assumed that dolomitic raw materials were used in the younger group. Their synthetic bodies are made with White of Bresse (a whitefiring local refractory clay), to which is added, in a proportion of one fifth, a blend composed of half limestone and half imported dolomite. This results in the following %: 80 White of Bresse + 10 lime + 10 dolomite, whereas the older group is made of 90 White of Bresse + 10 lime [55]. A detailed discussion of this specific dolomitic ingredient will be given in Section 4.3.

In the second quarter of the 19th century, at about the same time as Meillonnas’ group 2, the small Jussy manufactory, located not far from it on the southern shore of Lake Geneva, also produced a white earthenware rich in MgO (one marked piece from a private collection analyzed, Table 4). Here too, dolomitic ingredients must have been used. This is not surprising as the factory bought the recipe from the Swiss manufactory Robillard in Nyon situated on the opposite northern shore of the lake, see Section 4.3.

The XRD-determined phase association of the CaO-poor objects (Creil, Pont-aux-Choux, Sarreguemines) consists of cristobalite, mullite, and quartz. This shows, based on experimental firings [51,52] that temperatures above 1050 °C have been reached. Such high temperatures are in good agreement with the upper limit of ca. 1200–1250 °C for the first firing of these bodies (Table 1). The XRD phase associations of the CaO-rich white earthenwares are compatible with maximum firing temperatures of 800–900 respectively 900–1050 °C [16]. Such temperatures fall in the usual range of the French faïence kilns of the 18th century [59].

4.2.2. Glazes: Microstructures, Chemical Compositions and Technical Aspects

The glazes of two Pont-aux-Choux samples are macroscopically transparent, without any inclusions and consist mainly of SiO₂ (33–39 wt %) and PbO (50–57 wt %), Table 3. No reaction zones could be observed between the glazes and the bodies. Both Creil transparent and inclusion free glazes show severe burial leaching effects which made a chemical analysis very difficult. The result, SiO₂ 26 wt % and PbO 73 wt %, does evidently not correspond to the original composition and must be checked with objects from collections. There are no glaze-body reaction zones. Glazes on CaO-rich Lorraine bodies are macroscopically either transparent or opaque white, the latter containing many tiny tin oxide crystals (cassiterites SnO₂) and lack any reaction zone (Figure 10). Cassiterite-free glazes are much thinner than the other type. Regarding the conspicuous quartz grains in the glaze, see the discussion below. A glaze analysis of a single Niderviller white earthenware does not differ much from the Staffordshire and Wedgwood glazes if both main components are taken into account (SiO₂ 40 wt % and PbO 51 wt %). However, they have a higher Alkali content (2–5 wt %) [25]. A much younger sample from Niderviller and five from Saint-Clément have tin oxide opacified glazes, see Figure 10b [16]. Their composition is quite similar (

Table 5. Thicknesses and SEM-EDS analyses (wt %) of opaque (white) and transparent glazes of French CaO rich white earthenware.

	Niderviller		Saint-Clément
Date	ca. 1770	1793–1832	1755–ca. 1770/18th century
N samples	1	1	5
Reference	[16]	[25]	[16]
Thickness (µm)	100–140	80	300–500
Na2O	1.8	0.5	1.2–2.6
MgO	0.2	0.9	0.1–0.3
Al2O3	4.7	5.9	4.2–5.9
SiO2	48	40.4	45.2–47.6
Cl
K2O	2.3	0.3	2.0–3.2
SnO2	8.2		7.2–10.0
CaO	1	0.3	0.4–1.2
TiO2		0.1	0.2
Fe2O3	0.6	0.4	0.1–0.5
PbO	32.4	51.2	29.9–35.3

), with SnO₂ varying from 7 to 10 wt %, corresponding to traditional faïence glazes [6,59]. The use of an expensive ingredient such as tin is puzzling. This procedure could have been motivated by the fact that the transparent glaze was soft, easily corroded, and subject to crazing [23]. A rather hard tin-glaze, well adapted to a CaO-rich body such as that of traditional faïence and the new white earthenware—the coefficient of thermal expansion of both bodies matching those of lead alkali glazes [60,61]—made it possible to avoid these drawbacks.

The composition of Meillonnas’ glazes is not known, but a recipe for the younger group from 1817 is reported [56]: 200 p. (pounds) of calcined lead + 12 p. tin from Malacca + 160 p. sand from Nevers + 60 p. gravel ash + 24 p. soda from Alicante + 28 p. sea salt + 7 ounces of azur. All these, well mixed materials were put in the kiln, in the shape of a sugar loaf coated with wet sand, on a ceramic container. After fusion and cooling, a beautiful “crystal” is obtained that will be removed from the kiln with subsequent cleaning, pounding, sieving, and grinding.

4.3. Swiss White Earthenware

Between 1766 and 1803, white earthenware started to be produced in at least eight manufactories in Switzerland [19], see Figure 11, Figure 12 and Figure 13. In this country too, the archaeometric studies are only at the beginning, because very few pieces were analyzed and not from all manufactories: one unmarked bisque fragment from the production site in Fribourg [62,63], one marked Matzendorf fragment from archaeological excavations on a consumption site [19], 29 bisques and glazed fragments (only three stamped) found in archaeological excavations on the production site of Kilchberg-Schooren [64,65] and 39 marked objects (14 from consumption sites; i.e., archaeological excavations and 25 from museum and private collections) from the manufactories in Carouge and Nyon [37].

4.3.1. Body: Microstructures, Chemical Compositions and Technical Aspects

The bodies from the eastern manufactories (Fribourg, Matzendorf, Kilchberg-Schooren) of Switzerland differ significantly from those from the western ones (Carouge, Nyon).

Bodies from Fribourg (production period 1776–1810), Matzendorf (1800–1805) and from two older production periods of Kilchberg-Schooren (ca. 1800–1810, 1st quarter of 19th century [65]) have a CaO-content of 7 to 13 wt %, whereas the youngest Kilchberg-Schooren type pertains to the CaO-poor ware (

Table 6. Chemical compositions (XRF, Min.–Max.) of Swiss white earthenwares. Oxides in wt %, elements in ppm.

	Fribourg	Matzendorf	Kilchberg-Schooren			Nyon & Carouge
Date	1794–1810	1800–1806	ca. 1800–1810	1st Quarter of 19th Century	2nd Quarter of 19th Century	Nyon: 1790–1930, Carouge: 1803–1879				Carouge 1880–1928
						A	B	C	D
N analyses	1	1	17	9	1	7	8	8	4	11
Reference	[62]	[19]	[64]	[64]	[58]	[37]	[37]	[37]	[37]	[37]
SiO2	75.49	68.81	68.80–75.53	63.80–75.53	74.89	52.72–63.08	52.45–60.54	61.76–72.24	66.63–72.33	73.89–79.75
TiO2	0.62	0.54	0.16–0.26	0.29–0.49	1.42	0.50–0.84	0.63–0.87	10.49–0.78	0.61–0.76	0.32–0.65
Al2O3	9.21	8.55	6.29–9.25	10.91–16.21	17.64	10.74–19.05	11.93–16.89	10.63–16.51	11.12–15.93	16.35–23.50
Fe2O3	0.80	0.63	0.45–0.61	0.57–0.69	1.77	0.53–0.93	0.59–0.76	0.54–0.82	0.55–0.69	0.48–0.83
MnO	0.02	0.01	0.01	0.01	0.03
MgO	0.54	0.55	0.73–1.08	0.39–0.65	1.10	7.62–9.14	2.98–6.77	1.22–6.31	0.30–0.41	0.05–0.16
CaO	10.82	17.89	7.64–13.33	7.32–11.74	2.39	11.46–13.48	15.66–19.75	7.30–9.86	7.14–8.85	0.25–1.60
Na2O	0.42	0.98	0.35–0.99	0.39–0.95	0.26	0.39–1.31	0.34–0.91	0.74–1.27	0.96–1.93	0.28–1.08
K2O	0.81	0.86	1.05–1.98	1.02–1.53	0.70	1.68–2.24	1.08–1.90	1.48–2.33	1.53–1.94	0.64–1.89
P2O5	0.23	0.21	2.80–5.77	1.68–4.11	0.06	0.04–0.05	0.06–0.09	0.04–0.05	0.05–0.06	0.06–0.17
Ba	232	185	14–557	173–298	297	246–346	259–312	186–279	238–289	139–696
Cr	41	41	8–37	34–45	158	74–110	71–102	75–113	69–87	32–46
Cu	35	96	3–36	4–55	53	13–53	29–67	8–40	12–54	5–54
Nb	12	11	6–19	9–15	44	10–17	14–17	8–17	10–15	13–31
Ni	14	13	8–14	11–15	59	23–37	19–35	23–41	25–31	11–32
Pb	2151	2625	2100–10,365	1361–21,827	480	4666–24,511	12,183–23,565	4311–44,277	18,441–34,528	94–800
Rb	32	32	20–134	47–86	50	11–121	8–46	4–115	27–52	44–518
Sr	1164	1179	106–488	387–778	141	71–94	153–237	78–196	165–251	75–310
Y	19	23	11–66	24–38	16	8–61	14–50	6–55	9–55	11–33
Zn	45	222	21–62	32–69	42	32–61	21–46	31–62	21–42	28–190
Zr	140	146	98–155	105–232	182	104–250	163–262	51–217	80–223	121–188

). Obviously, around 1825/1830, the paste recipe of this manufactory changed from CaO-rich to -poor.

Both CaO-rich Kilchberg-Schooren types usually plot in many binary correlation diagrams into two chemically distinct groups (Figure 14).

The objects of the older period are MgO-richer, but TiO₂- and Sr-poorer (one exception) than the slightly younger pieces (Table 6, Figure 14a,d). This dichotomy is also nicely expressed in the Sr-CaO diagram, which suggests that during both time spans two recipes were used with at least two CaO raw materials showing different Sr/Ca ratios (Figure 14d).

The P₂O₅ content of the samples from this manufactory markedly exceeds the maximum of 0.7 wt % found in the European white earthenware analyzed so far (Table 2, Table 4 and Table 6; Figure 14c). Postfiring contamination (e.g., burial contamination through migrating P-rich solutions) and admixing of P-rich material (e.g., calcined bones) before firing are two processes normally advocated to explain such phosphorous anomalies (see literature in [66]). But such unusual P-concentrations could also be due to phosphoritic elements present ab initium in the clays [67]. In the present case, burial contamination can be ruled out because two foreign sherds from the same archaeological level have no such P-anomaly. The studied bodies contain many small grains with an internal microstructure typical for calcined bones (Figure 15 and Figure 16a). The chemical composition of these grains, as determined by SEM-EDS, fits fairly well with a Ca-apatite, the inorganic mineral phase of bones.

This bone powder was added as an additional flux and represents a characteristic of the Kilchberg-Schooren production. One may wonder where the manufactory got this peculiar idea. Three bisques of the older production period carry the stamped mark of Johann Wilhelm Spengler, son of the first director of the porcelain manufactory Kilchberg-Schooren, who visited England (Derby) in 1790 and brought new forms and the secret of the white earthenware to Kilchberg [68]. Possibly he copied the idea of a phosphorous flux from the English bone china industry, though it is very astonishing that in the first years he produced a CaO-rich white earthenware in the French manner and not a CaO-poor creamware in the English tradition.

In addition to bone, Pb-bearing frit fragments were also added to the Kilchberg-Schooren paste mixture, as confirmed by SEM-BSE and EDS analyses [64] and the high Pb content of the bodies, which exceeds in some cases 1–2 wt % (Table 6).

The CaO-poor bisque of the 2nd quarter of the 19th century is chemically markedly different than the English and French CaO-poor samples of Table 2, as documented by its higher titania, magnesia and lime, and lower alumina.

The Fribourg and Matzendorf samples differ in some major and minor oxides and trace elements from the Kilchberg-Schooren products and can therefore be easily distinguished using simple binary diagrams (Figure 14). Their high Sr must be a specificity of the carbonate admixture, whose origin cannot yet be determined.

The SEM-BSE image of the Matzendorf specimen differs from that of Kilchberg-Schooren (Figure 16). Although comparable proportions of splintered quartz fragments can be seen in both, the quantity of matrix in the former is higher, which is reflected in a higher CaO content (Figure 14c). The angular outline of the quartz grains is a clear indication that quartz was added as a ground powder. Bone admixture is not mentioned in the Matzendorf manuscript recipe, but some bone chips could be detected. The SEM analysis showed no evidence of Pb frits, which fits well with the low Pb content of ca. 0.3 wt %, most probably caused by contamination by the aqueous glaze suspension before firing and/or infiltration of Pb vapors or of the molten glaze during the firings (see discussion above). The microstructure of the Matzendorf sample is an artificial mixture of white firing clay, calcined quartz pebbles and chalk in a ratio of 2:1:1 as reported in [35]. According to the recipe and the BSE images of Matzendorf, the calcium could have been added as quicklime CaO or slaked lime Ca(OH)₂ and not as the usual powdered chalk CaCO₃. The clay was imported from Southern Germany over ca. 100 km; i.e., from the village of Heimbach north of the town Freiburg i. Breisgau while the quartz pebbles were collected nearby, in the river Emme. Where the CaO raw material comes from is not known. The recipe book lists further mixtures, including some with only local raw materials. The variety of these recipes proves that Matzendorf constantly adapted their pastes. In addition to powdered bisque (grog) lead-free frits are also mentioned as compounding ingredients.

The analyses of white earthenwares from Western Switzerland led to the astonishing result that in Nyon, for more than a hundred years, from 1790 to 1930, only CaO- & MgO-rich pastes were used (Table 6). This also applies to the products of the Carouge factories for the period 1803 to 1880. In both sites, the English and French technical progress has therefore been completely ignored. In 1880 however, a radical break with the past occurred in Carouge, where CaO-poor (0.2–1.6 wt %), SiO₂- (70–79 wt %) and Al₂O₃-rich (16–23 wt %) crockery was produced until 1936.

The microstructure of the eleven CaO-poor samples is similar (Figure 17). It shows a very fine-grained matrix composed of tiny quartz granules, former phyllosilicates and few TiO₂ phases (rutile, anatase). The matrix contains four types of non-plastic inclusions: (1) angular SiO₂ (quartz according to the XRD diffractograms) grains with maximum sizes of 100 µm. Their appearance is homogeneous without evidence of internal fracturing; (2) rare type A grains with angular contours and similar dimensions as the quartzes, showing in some cases partial melting at the borders. They are filled with small sticks or needles of a whitish color under the SEM, most likely mullite crystals embedded in a vitreous phase (Figure 17c); (3) fairly frequent grains, type B, with micromorphological aspects typical for kaolinite, with a maximum diameter of 30 µm, often showing signs of melting in contact with the matrix. Their outline goes from angular to rounded (Figure 17d); and (4) rare pores (maximum diameter 70 µm) with a compact and fine halo, typical of former calcite or limestone grains decomposed during firing.

The body is, according to the microstructural analysis, an artificial one, made by a mixture of white firing clays, quartzes (probably previously calcined) and other ingredients such as inclusion types A and B. The chemical composition of the matrix (without added non plastic inclusions) is very similar to that of the white firing clays of the Westerwald [16]. The angular contours of the quartz and the type A and B grains are clear evidence that these particles were added to the clays after thorough grinding.

The type A inclusions have a microscopic appearance and a chemical composition very close to that of similar type A inclusions of the Wedgwood tableware (see before). These were interpreted as SiO₂-rich kaolinitic clays (ballclays) or mixtures of them with Cornwall Cornish stone, a kaolinitized peraluminous granite rich in feldspar [25]. For the Carouge tableware, the angular contour and their homogeneous nature indicate rather that they are grains obtained by the grinding of a feldspar rich granite of the Cornish stone type. The ceramic firing melted the feldspars and mullite crystallized out of the liquid on cooling.

Type B inclusions very often look like large grains of phyllosilicates cut parallel to their crystallographic axis c. The chemical composition brings them close to the type B inclusions of Wedgwood’s crockery, although their microscopic aspect is different. The latter were interpreted as an aluminous ball clay admixture [25]. In the case of the studied Swiss sample, they are rather former crystals of kaolinite, as evidenced either by their characteristic (relic) appearance or their chemical composition, which is close to that of kaolinites. The potassium and iron content indicates that the original minerals, prior to their transformation into kaolinites, were micas.

In conclusion, the chemical and microstructural analyses of the youngest Carouge tableware suggest that their paste was probably a mixture of crushed quartz (sand), white firing clay, kaolin, and subordinate carbonate rock (Table 6, Figure 17). This fits well with the recipe of 1880 (in wt %): 20 clay (a blend of four clay types) + 41 kaolin (two types) + 37 sand (two types) + 2 chalk [36]. Due to the kaolin admixture they are no longer creamwares, but transitional types to the Queen’s wares without kalifeldspar, see also Figure 7d.

The contact between the ceramic body and the glaze is sharp and often highlighted by an extremely thin layer of microcrystals. Precise chemical analysis of these crystals is very difficult because of their small size. They seem to be rich in Si, Al, Pb, and Ca. These crystals have their origin in the reactions between the glaze and the ceramic body that took place during the second (glaze) firing. Other important effects of these reactions are clearly visible, for example on quartz grains. In the ceramic body, their shape is angular, but round when immerged in the glaze at the contact body/glaze (Figure 17b). Evidently, these minerals were incorporated in the liquid glaze during firing. Where the grains were embedded, the matrix has also disappeared. It can be estimated that at least 10 microns of the body thickness were digested by these processes. Moreover, qualitative measurements showed a significant lead infiltration into the ceramic body, as still 0.4 wt % PbO was detected in the body at a distance of 200 µm from the contact with the glaze. Close to the contact and over a distance of ca. 40 microns, the microstructure of the body is very different from the rest, showing lower porosity and signs of partial melting. Both are induced by Pb, which moved from the liquid glaze into the body during firing. According to [36], the temperature of the first (bisque) firing was at 1280 °C while that of the second firing reached some 100 °C lower temperatures of about 1180 °C. This is corroborated by the presence of cristobalite and mullite in the diffractograms of the majority of the samples (n = 8), which indicate that they were fired above 1050 °C, by reference to the results of experimental firings of kaolinitic clays [51,52]. Two samples had to experience temperatures between 1050 and 950 °C and one a temperature close to 950 °C.

The chemical study of 28 CaO- & MgO-rich samples from western Switzerland (Carouge, Nyon) revealed an unexpected complexity, as four subgroups were found (Figure 18a). The first, group A, brings together all analyses with a MgO/CaO ratio corresponding to an exclusively addition of a dolomite rock or the mineral dolomite (Figure 4 and related discussion). The richness in CaO and MgO of the bodies of the second, group B, implies important additions of two carbonate ingredients: dolomite + limestone or a dolomitic limestone alone. The third, group C, samples have variable MgO/CaO ratios, reflecting additions of dolomite + limestone or dolomitic limestone alone, but at doses much lower than the second group. The objects of the fourth, group D, show high CaO and very low MgO concentrations, fitting well with a limestone addition without dolomite.

This dual nature of the carbonatic ingredients finds its reflection in the distribution of the chemical elements as shown in the mappings. In the type B body of Figure 19, calcium appears to be related to small and distinct Mg-rich grains (=former dolomites), but few grains show much higher Ca-concentrations without any Mg-relation and are therefore former calcites.

All western Swiss white earthenwares, except the youngest sample N10 of the years 1920–1930 (what probably shows the effects of the modern legislation in health matters), are rich in fragments of a lead frit, which are lacking in the bodies of the CaO-poor ware (Figure 19a and Figure 20). It is therefore not surprising that the Pb concentrations reach almost 5 wt % (Table 6). Some of them contain small amounts of cassiterite crystals. The maximum diameter of these grains is less than 100 µm. They have a subangular outline, with a very sharp contact with the surrounding matrix. The SnO₂-free frits consist of two phases, a lead-rich glass and SiO₂ crystals. The latter are either irregularly shaped cores, or needles and rods surrounding these mostly roundish cores. These are interpreted to be relic quartzes having survived a complete fusion during the fritting process. The needles or rods crystallized from the PbO-SiO₂-rich liquid at the T climax and on the retrograde path during cooling of the frit. They could be tridymite or cristobalite. The amounts of vitreous phase versus SiO₂-phases vary enormously, ranging from grains with a dominant glassy phase (Figure 20a) to grains consisting almost exclusively of SiO₂ neoformations (Figure 20b), as those in the Lorraine samples.

The sharp boundary with the matrix shows that the glassy portions of the frits were not liquefied enough during the bisque and the glaze firings of the objects, and that mechanical infiltration of such a melt into the matrix could have occurred. Lead has diffused over a distance of ca. 10 µm into the matrix, as evidenced by profile analyses. Such behavior corresponds well with the inferred maximum temperatures of 1050 °C for the firings (see below).

The microstructure of the bodies type A is very characteristic. It shows, apart from the mentioned ingredients (quartz, frit), many black pores with rectangular to rhombohedral contours and maximum diameters of 15 µm (Figure 21a). Such voids also appear in body types B and C (Figure 21c). Some of these pores are filled with a cluster of tiny crystals. The SEM-EDS analyses indicate high Mg concentrations in these voids (Figure 19). The typical shape and the chemical composition indicate that these pores most likely correspond to former, ground dolomite grains that were dissociated during firing. The breakdown of the dolomite grains was not a simple process, as shown by a coarse grain with empty core, but still rich in magnesium, surrounded by a halo of calcium. Perhaps this element migrated towards the edge of the former dolomite rhombohedron or the observed feature is a primary one.

These four types of carbonate additions, either as minerals or rocks, were all reported in the recipes of Antoine Louis Baylon [32]. The chemical composition and the microscopic appearance of the objects analyzed reveal artificial paste mixtures similar to those described by this potter with the following ingredients: sand (from Cruseille) + white firing clay from Germany (Cologne) + Pb frit + chalk + Morez white + gypsum. The Morez white is most probably the source of the magnesium, but neither its exact nature (Pure dolomite? Rock or mineral?) nor its origin are known. According to Baylon, this ingredient replaces the chalk from France (Champagne) of earlier recipes, from which one can deduce that it must be a local carbonate. Importing chalk from France (Champagne) was an expensive operation and the financial aspect certainly played a crucial role. Replacing it by a local raw material was certainly a welcome pecuniary solution. There is however most likely an additional (technical) reason for this choice. The mineral dolomite dissociates at temperatures of 100 °C lower than those of calcite, as has been shown by many experimental firings of dolomitic clays (references in [37]). For a dolomitic paste, the appearance of firing minerals such as diopside, gehlenite, and plagioclase as well as the reorganization of the microstructure will therefore require a temperature of 100 °C less than in the case of a calcitic paste—a significant thermal, fuel, and financial advantage. It may be recalled that French and German faïence factories located on Triassic terrains used routinely dolomitic; i.e., magnesian marls [69,70].

Pb frits were added in many Central European pottery centers to CaO-rich pastes. This practice comes from the manufactories in Lorraine, who used them as an additional flux to the main carbonate flux (chalk from Champagne) to promote even better reactions during firing and to lower the firing temperatures. The French technical treatises of the first half of the 19th century extensively detail the recipes of contemporary Al-rich white earthenwares, but deliver very little useful information and no recipe for the CaO-rich terre de pipe (see above). This is not very surprising because at that time, terre de pipe was no longer popular among the ceramic engineers, although it continued to be produced in many places, as exemplified for instance in Nyon. Fortunately, recipes have survived in handwritten sources that mention alkaline or lead-bearing frits [16] (Table 1). Both are detailed in Antoine Louis Baylon’s workbook, but the chemical composition of the analyzed frits, all of the Si-Pb-Al-K-Ca type, does not match the recipes of A. L. Baylon. The cassiterite-bearing frit was also not mentioned by A. L. Baylon. This is very likely a waste product of the stanniferous glaze originated during the production of faïence, another ceramic type still produced in Carouge and Nyon alongside with the white earthenwares.

The 28 western CaO- & MgO-rich samples are easily distinguishable from the eastern ones, for instance by the much higher chromium and nickel content. This is a trait they probably inherited from the ophiolithic chromites in the Cruseille sand (Figure 18b). On the other hand, there are hardly any important chemical and microstructural differences with Jussy’s products. This is not surprising because the secret recipe of Nyon was sold in 1824 by Robillard to this manufactory [37].

The vast majority of the eastern and western Swiss products were fired at temperatures between 950 and 1050 °C, which most probably corresponds to those of the first (bisque) firing [19,37,64].

4.3.2. Glazes: Microstructures, Chemical Compositions and Technical Aspects

The transparent, colorless and maximum 220 µm thick Al₂O₃-rich PbO-SiO₂ glaze of the studied Matzendorf sample is, except for a few needle-shaped crystals, inclusion-free (Figure 22a,b) [19]. The high concentration of SiO₂ and PbO demonstrate the use of quartz and a lead compound. Salt can be recognized as a further ingredient on the basis of Na₂O and chlorine values (

Table 7. Thicknesses and SEM-EDS analyses (wt %) of Swiss and Italian white earthenware glazes.

	Matzendorf	Kilchberg-Schooren		Nyon & Carouge				Carouge	Torino
Date	1800–1806	ca. 1800–1810	1st quarter of 19th century	Nyon: 1790–1930, Carouge: 1803–1879				1880–1928	1824–1846
				A	B	C	D
N samples	1	3	1	7	8	8	4	11	1
Reference	[19]	[64]	[64]	[37]	[37]	[37]	[37]	[37]	[37]
Thickness (µm)	220			57–117	35–99	21–82	42–84	28–141	75
Na2O	3.1	0.7–1.1	1.2	0.3–1.8	0.2–1.4	0.2–2.7	0.2–0.7	0.5–1.7	1.3
MgO	1.3	0.4–0.6	0.7	0.3–0.5	0.3–1.4	0.2–0.4	0.2	0.1–0.8
Al2O3	6	4.4–5.7	5.7	2.6–5.7	3.1–4.7	3.1–9.5	3.3–5.0	9.6–14.8	5.1
SiO2	45.2	48.2–51.3	48.3	28.2–55.9	37.0–48.4	38.4–66.6	35.7–45.0	49.9–60.4	43.9
Cl	0.2			0.4	0.4–0.5	0.4–0.8	0.5
K2O	1.5	2.2–3.0	2.9	0.9–3.8	1.4–3.7	0.7–5.2	1.2–1.9	2.2–3.2	1.9
SnO2		0.7–3.2	0.3	(2.2)
CaO	3.1	1.5–1.6	3.2	0.2–2.2	0.3–1.3	0.1–0.5	0.8–1.9	3.1–7.7	2
TiO2				0.4	0.1	0.1–0.5		0.2
Fe2O3	0.9	0.7–0.9	0.6	0.4–0.6	0.4–0.6	0.4–0.6	0.5–0.6	0.3–2.1	0.4
PbO	38.8	33.1–40.0	37.1	30.7–66.4	40.0–56.0	13.8–52.9	45.1–56.9	16.6–25.2	45.4

). These ingredients correspond to those of the recipes as mentioned in the compendium called Arkanum [35]. There are three recipes for the period 1800–1806. The one dating from 1805 only lists four ingredients (in %): 37 sand + 40 minium + 21 salt + 2 alum. Potassia and alumina amounts seem to match the alum addition mentioned in the Arkanum. However, 1.5 wt % K₂O from alum would give max. 1.6 wt % Al₂O₃ from the alum. The high alumina must therefore be explained otherwise. Further, the magnesium and calcium contents cannot be explained with the pure raw materials mentioned in Matzendorf’s recipe. It must therefore be assumed that the raw materials were not pure or that not all the ingredients are listed in the recipe. In addition, as already stated, chemical exchange process could have taken place between the glaze and the body, affecting the entire cross section of the glaze. The first case is probably the most likely, especially considering quartz sand, which may well have contained some percent kaolinite (Al supplier) and carbonate (Ca and Mg supplier). From these imponderables and the fact that in the glaze firing a large portion of sodium evaporates, the original composition of the glaze-raw material mixture cannot be reconstructed. A quantitative comparison with the Arkanum is therefore impossible.

The glazes of the Kilchberg-Schooren CaO-rich products are clearly Na₂O- and CaO-poorer, but SiO₂- and K₂O-richer than the Matzendorf glaze (Table 7). They are characterized by a significant tin oxide content that is absent in Matzendorf. Lorraine CaO-rich white earthenwares have either a white opaque lead glaze with significantly higher SnO₂ (8–10 wt %) or a colorless transparent Pb glaze (see above and Table 5). However, the K₂O and CaO contents of the latter do not reach 1 wt % and are thus far below to that of the Eastern Swiss white earthenware.

The chemical analyses of glazes the Western (Carouge, Nyon) Swiss CaO- & MgO-rich white earthenware are compatible with the ingredients of A. L. Baylon’s recipes, either sand or pebbles + salt + minium or litharge + potash ± alum [32] (Table 3), taking into account: (1) the decomposition of salt during the sintering of the mixture constituting the glaze, followed by the evaporation of sodium and chlorine during this process and then also during the firing of the glaze; and (2) the contamination of the glaze by reactions between it and the ceramic body (see below).

Under the microscope, the glazes have a very clear appearance, except NYO 10, which stands out for its richness in quartz grains (Figure 17c,d). For the latter, the potters added crushed quartz powder to the raw glaze. The glaze recipe of the youngest products (1920–1923) is therefore very different from the 19th century objects that did not use any secondary quartz powder. The angular shape of these quartzes suggests that they have not or only slightly reacted with the liquid glaze during the second firing of the ceramic object. Adding powdered quartz to the raw glaze was a common practice in 18th century French, German, and Swiss faïence manufactories [63,64,71,72,73,74,75,76,77,78]. The purpose of such a technique is unclear: increasing of the translucency of the glaze [79]? Compaction of the glaze powder [61]? No old or modern technical manual refers to this type of addition [72]. The fact that fusion of pure quartz gives a very viscous liquid could indicate a solution: the potters added crushed quartz simply to increase the viscosity of the molten glaze to prevent it from flowing too easily from the ceramic surface.

The contact glaze-ceramic body is sharp. But significant chemical reactions must have taken place between both, because granules of the body surrounded by the vitreous phase, visibly corroded quartz grains, and crystals of neoformation are observed in the glaze close to the contact. Such phenomena were also reported from the Matzendorf sample where the body lost at least 20 microns of its substance on a vertical section. For the Carouge and Nyon wares, it is probable that the glaze absorbed about 30 microns of the body on a vertical section, if the upper limit of the small crystals in the glaze is taken as representing the original surface of the body. Experimental firings have shown that Al of a pre-fired kaolinitic substrate migrates over a distance of 100–200 µm into a lead glaze [54]. Because the maximum thickness of the analyzed glazes was lower than 120 µm, a large part, if not all, of Al and probably also other chemical elements, could have been imported into the glaze by these high temperature reactions. Therefore, the chemical composition of the glazes no longer reflects the initial recipes. These complex digestion-diffusion reactions have been discussed in detail [53,54].

Even if the effects of the chemical reactions between the body and the glaze during firing are taken into account, it is to be noted that the glazes of the Al-rich white earthenwares from Carouge are much richer in Al₂O₃, K₂O and CaO than the CaO- & MgO-rich white earthenwares from Carouge and Nyon, and also than the Al₂O₃-rich tableware of Pont-aux-Choux and Wedgwood (Table 3). Such a richness cannot be completely related to the chemical interactions body-glaze, but finds its explanation in the glaze recipe. This involves inter alia kaolin and a frit, which is composed of feldspar + kaolin + chalk + other ingredients. The high levels of Al₂O₃ are therefore due to kaolin and feldspar, those of K₂O to feldspar and those of CaO to chalk. The lower values of these oxides in the CaO- & MgO-rich white earthenwares from Carouge and Nyon are easily explained by the absence of these three ingredients in their glazes recipes.

4.4. Slovenian and Italian White Earthenware

40 stamped and 46 non stamped white earthenware objects from the National Museum of Slovenia collections, originating from four Slovenian and three Italian manufactories (Figure 23), were recently studied with non-destructive analytical methods [45].

Table 8. Chemical compositions (Min.–Max.) of Slovenian and Italian white earthenwares. Oxides and chlorine in wt %, other elements in ppm.

	Ljubliana			Nemski dol	Kamnik	Trieste		Pordenone	Torino
Date	(Zois) 1795–1816		(Wasser) 1817–1831	1817–1928	(Konjsek) 1855–?	(Lorenzini) 1776–1798	(Santini) 1784–1813	(Galvani) 1797–1855	1824–1846
N analyses	12	2	7	1	6	4	5	3	1
Method	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	PIXE-PIGE	XRF
Reference	[45]	[45]	[45]	[45]	[45]	[45]	[45]	[45]	[37]
	CaO-rich	CaO-poor
SiO2	36.8–68.0	41.1–43.4	52.0–62.5	56.9	56.1–69.3	54.8–71.7	61.8–74.0	53.1–62.0	58.41
TiO2	0.11–0.23	0.03–0.05	0.12–0.44	0.13	0.03–0.12	0.13–0.17	0.08–0.13	0.12–0.15	0.31
Al2O3	12.0–22.9	4.72–10.0	10.6–16.5	15.5	9.71–20.4	13.1–28.2	5.84–15.9	11.1–21.9	11.70
Fe2O3	0.33–0.81	0.21–0.23	0.21–0.43	0.19	0.16–0.37	0.47–0.61	0.50–0.86	0.39–0.79	0.59
MnO	0.01–0.02		0.01–0.05		0.01–0.02	0.01	0.01	0.01
MgO	0.75–4.93	0.48–0.73	1.44–4.01	1.96	0.26–0.52	0.81–2.17	0.63–1.74	1.24–1.36	10.97
CaO	6.58–32.6	0.97–0.98	13.1–21.9	16.5	0.97–3.26	5.88–7.30	9.51–15.0	9.53–12.5	15.63
Na2O	0.39–0.54	0.18–0.20	1.01–2.09	1.62	2.47–3.95	0.18–0.39	0.43–0.90	0.22–0.59	0.63
K2O	0.76–5.49	1.16–1.47	2.00–5.84	4.69	1.39–2.47	2.36–4.85	1.45–2.41	1.94–3.98	0.63
P2O5									0.10
SO3	0.47–9.61		0.63–3.62	0.67	1.50–2.02	1.02–2.60	1.00–5.94	4.69–7.27
Cl	0.06–0.62	0.91–0.97	0.06–0.31	0.30	0.14–0.90	0.13–0.23	0.17–0.32	0.19–0.36
Ba									191
Cr									58
Cu	1050					400–580			22
Nb									7
Ni									45
Pb	5477–225,577	412,165–443,727	8344–65,073	14,667	31,933–235,788	27,292–107,682	5384–55,884	3527–99,328	9242
Rb									22
Sr									156
Y									12
Zn	56–440		100–500		67	100–230	100–200	100–200	83
Zr									79

reports only the results of the stamped ware. These are joined by an XRF and SEM-EDS analysis of one object from the Turin manufactory Dortu, Richard & C^ie [37].

4.4.1. Body: Microstructures, Chemical Compositions and Technical Aspects

For the production of the Slovenian pastes, white-firing clays of local or Italian origin have been mixed with chalk [45]. However, the PIXE-PIGE results in Table 8 show high SO₃ values (up to ca. 10 wt %), which were not discussed by the authors. They could be explained with gypsum admixtures providing they are not analytical errors. The chlorine values were not further addressed either. They could result from a contamination of the analyzed surfaces and/or a NaCl admixture in the recipe. Most Slovenian bodies belong to the CaO-rich group, and only two were made after a CaO-poor recipe, not noted by the authors. The most surprising, however, are the exceptionally high Pb contents, which reach up to 44 wt %. The authors found a PbO concentration of 0.12 wt % below the surface, which they attributed to lead diffusion. However, such a contamination process was not discussed in detail and the measured Pb values (transformed into PbO: max. 47.8 wt %) in the bodies far exceed these 0.12 wt % PbO, a fact that was not explained. In addition, prior to the chemical grouping, if the lead is indeed due to contamination, the analyses should have been converted to lead-free. As this did not happen, all the classifications and attributions of [45] must be reconsidered. If these high Pb values are really those of the body and not of both body and glaze, they could be explained not only by contamination processes (see discussion above), but also by the addition of Pb frits, as was the case in the Lorraine factories. A SEM-EDS study would clarify this.

For the analyses of the Italian white earthenwares from Pordenone and Trieste the same restrictive remarks apply as for the Slovenian. All have CaO-rich bodies. Their high Pb is probably indicative of the use of a Pb frit in the recipe of the three manufactories. In the case of Trieste, the older ones (Lorenzini production period) appear to be distinguished from the younger products (Santini period) by lower CaO and higher K₂O values. The high SO₃ concentrations of the three objects from the Galvani factory in Pordenone differentiate them from those of Trieste. The body of the sample from the Torino factory corresponds to a dolomitic ingredient, as shown in Figure 18a. It occupies the highest position on the red line, which indicates the highest dose by far of dolomite (rock or mineral) of all the white earthenware analyzed. It differs from the Swiss objects by a higher content of quartz inclusion and a higher porosity (Figure 21d). Regarding the Cr- and Ni-concentrations, Torino’s sample is not part of the Swiss groups (Figure 18b). This is hardly surprising, since Frédéric Dortu, son of the director of the manufactory Carouge and founder of Torino, used only local resources for his recipe in Torino, dated 18 August 1829 (no unit given): 450 Mondovi earth + 300 Calcium carbonate + 400 frit [37]. The frit was made with 150 sand from Castelamont + 18 litharge + 6 potassia. The recipe of the paste and the frit can be found, but no information is given on the unit of measurement or where these local ingredients came from. At present, nothing is known about the Mondovi earth. It could be a white firing Al rich clay like the one from Cologne. Dortu also gives no information about the use of a quartz ingredient (sand) in his recipe. The fact that such an addition took place is confirmed by the BSE image (Figure 21d). And the calcium carbonate is perhaps a pure limestone as suggested by the recipe. However, according to the chemical analysis, it would rather be a dolomitic rock. It can be noted that for at least this piece the recipe of 1829 does not agree with the analyses. As only one object from Torino was examined so far, no firm conclusions on the entire production can be made.

The XRD phase association of the Torino sample is compatible with maximum firing temperatures in the interval 900–1050 °C [37].

4.4.2. Glazes: Microstructures, Chemical Compositions and Technical Aspects

Frédéric Dortu revealed in his recipe of 1829 no information on the glaze production. Clues to the glaze recipe must therefore rely on microscopical and chemical analyses. The glaze of the object from the Torino manufactory does not differ chemically and microstructurally from those of Carouge or Nyon. Accordingly, the statements made in this context also apply to the analyzed specimen.

5. Conclusions

• White earthenware was produced in Europe in enormous quantities in small factories as well as in large industries from the middle of the 18th century onwards. However, archaeometric studies are just at the beginning.
• Traditional potters, for example those of the tin-glazed faïence manufactories, used successful recipes for the bodies and the glaze for decades into the 19th century. In contrast, the recipe books of the white earthenware potters and the archaeometric analyses prove that these potters constantly changed and adapted the recipes. The detection and pursuit of this technical evolution is an exciting archaeometric research field.
• The white earthenware pastes or their fired counterpart, the bodies, are artificial mixtures. Two technical traditions can be recognized in Europe. In England, only CaO-poor clays were used. These were mixed with ground flint (creamware) and later additionally with kaolinite and Cornish stone (Queen’s ware). In France, CaO-poor as well as CaO-rich pastes were used. The first type, a blend of a CaO-poor clay with grog, is the one applied in Paris (e.g., Pont-aux-Choux manufactory). CaO-poor clays were also used in Lorraine, but mixed with marls or limestones, fritted quartz pebbles and lead frits (terre de pipe). Over time, these recipes were replaced by the more successful English one. First by a paste made of CaO-poor clays mixed with quartz powder leading to the so-called cailloutage, and later with additionally kaolinite and other raw materials rich in potassium feldspars. This resulted in a product called faïence opaque, porcelaine opaque etc.
• In addition to CaO-rich ingredients (calcite, limestone, marl), glassy frits and kalifeldspar-rich raw materials, other fluxes such as bone ash or dolomite were also experimented. Further flux types are lime, chalk, giobertite (magnesite MgCO₃), pegmatite, glass, and fusible sand [23].
• Starting from France, the CaO-rich technique spread throughout Central Europe. Some manufactories retained this recipe until the beginning of the 20th century, obviously for financial reasons, while others switched in the first half of the 19th century to the more shock resistant CaO-poor pastes.
• In Western Switzerland (Carouge, Nyon) and in the neighboring French (Jussy, Meillonnas) and Italian (Torino) manufactories, dolomitic raw materials were used, probably because of the lower firing temperatures (cost issue). The geographic and temporal extension of this regional feature has yet to be clarified.
• White earthenware is fired twice, first in the bisque, then in the glaze firing. In CaO-rich products, the bisque firing temperature is lower than the glaze firing temperature, but higher in CaO-poor specimens. During the second firing, important physico-chemical reactions between the aggressive liquid lead glaze and the underlying body profoundly change the initial chemical composition of the raw glaze. Therefore, in many cases, the original glaze recipe cannot be calculated from the oxide concentrations.