M. Járó - L. Költő szerk.: Archaeometrical research in Hungary (Budapest, 1988)

Analysis - ZIMMER Károly: Spectrochemical investigation and classification of Hungarian glass finds

Sb. As trace elements, Ag, B, Ti and V were found, and very low amounts of Co, Cr and Ni could also be detected. The experimental conditions of the measurements for the determination of the quantitative composition of glasses were described in refs. [2, 3]; this paper reiterates the most important points only. Standard samples were prepared synthetically. For this process, the main components of the original material (Si0

2

, Na

2

0, K

2

0, CaO, MgO, A1

2

0

3

) were determined by thermometric rapid analysis [4]. The model matrix was diluted with a mixture of Li

2 C0

3 and graphite powder in order to improve evaporation and stabilize the discharge [5]. The measurements were performed on a PCS 2 plane grating spectrograph (Zeiss, Jena) using d. c. arc excitation. The line densities were measured by a modified microdensitometer of GîI-MFKI type with digital output [6]. The measurements were evaluated by computer using a specially developed program [7]. Before the discussion of analytical results, some problems had to be solved. With the matrices applied, no significant matrix effect could be found. The investigation of evaporation processes has shown that the most appropriate reference element for the volatile Cu, Pb and Sn elements is Ga, whereas for the medium and hardly volatile Fe, Co, Ni and Ti elements it is Pd. The application of the chosen Gil— MFKI microdensitometer involved a number of advantages [8]. The possibility for the reliable determination of high densities (up to S = 4) resulted in an analytical curve linear in a wide concentration region. The significant reduction of stray light at high densities leads to similar results: both the relative precision and the accuracy of the method is improved. Three methods of evaluation were used for constructing the calibration curve [1]: (i) with reference element, (ii) without reference element, (iii) using the background as reference density. The method which proved to be the best in the optimization of spectrographic calibration and evaluation was applied in the particular cases for the given elements [9,101 Colour classes. By means of the statistical analysis of the mean values characteristic of the various colour classes it was possible to reveal which elements and which concentrations show major differences for the various classes and which elements may play a significant role in colour changes. The results ob ained are shown in Table 1 for the colour classes. From the data the following conclusions can be drawn

. The white class is characterized by a high Sb content. According to some authors the white colour is due to the decolouring effect of antimony. The violet class has a high Mn content. The blueish or reddish shades are caused by the variation of the Cu content. The amount of Pb and Sn is in good correlation with the copper content. Literature sources attribute the variation of colour from pink to dark violet to the variation of Mn content. The claret and red classes are characterized by a high Cu and Pb content. Within the classes, the Sb and Fe and, in part, the Ti content varies, which is always higher in the case of lead containing samples. In the case of the reddish violet sample the Mn content is higher and the Pb content is somewhat lower. In the red colour class the Cu content of the samples is markedly higher. With Cu as colourant, it should be pointed out that red is obtained if the copper content is accompanied by a higher Sn content, this is the case with the claret and red classes. On the other hand, blue is obtained if the Sn content is low (blue class). The phenomenon may be interpreted in a way that the addition of tin suppresses the oxidation of copper present in the form of Cu

2

0 [11]. The shade of colour can be reached by changing the Cu:Fe ratio of the additive [12]. This explains the significantly higher iron content of some samples.