Paul-Louis Cyffle's (1724-1806) search for porcelain

Seven unglazed figurines from private collections, attributed to Cyffle's Terre de Lorraine manufacture in Luneville (1766-1780), were subjected to porosity and scanning electron microscopic analyses to determine bulk element compositions and the composition and nature of their constituent phases. One figurine bears the mark CYFFLE A. LUNEVILLE, three the mark TERRE DE LORRAINE. The studied samples pertain to four compositional groupings: (1) Two porous (18-20 % water adsorption W.A.) bodies containing milled quartz-rich frit + anhydrite (former gypsum) + phyllosilicate + Ca-rich matrix; (2) A soft-paste (artificial) porcelain body (10% W. A.), containing quartz + calcic plagioclase (An (sub 88-95) ) + glassy matrix. The latter is inferred to derive from a former frit. A coronitic, amorphous (as revealed by electron backscattered diffraction analysis) reaction rim is visible around the quartzes. The K-rich and Na-poor composition of the frit is best explained as a mixture of potassium nitrate, alum, calcined gypsum, sand, and moderate amounts of salt and soda; (3) A porous (23% W. A.), hybrid porcelain body with finely milled particles of quartz, mullite-bearing hard-paste porcelain, Na-Ca-siliceous glass and metakaolinite; (4) Three hard-paste porcelain bodies, some with relict quartz, andesine plagioclase (An (sub 37-45) ), pseudomorphs of kaolinite and the liquidus phase mullite in a glassy matrix. Well-fired figurines have no W. A. due to the pervasive former melt phase, underfired figurines 7%. These wares can contain small amounts of lead (1.8 wt% PbO) and SO (sub 3) (0.6 wt%), suggesting the use of lead frit and gypsum. The diversity of Cyffle's production is now better recognized. His trial-and-error experiments made use of a remarkably wide range of paste mixtures, with porcelain bodies in the French (soft-paste) and the German (hard-paste) tradition.