Making Wall Paintings: An Attempt to Reproduce the Painting Techniques of Bronze Age Thera

The objective of this endeavour is to understand the practical difficulties, and to apply the solutions which are imposed by the materials and the technique themselves. The modern samples were studied by X-Ray diffraction analysis, and polished sections of them were examined under an optical microscope and compared with the archaeological ones.

INTRODUCTION 

Over the last twenty years there has been much discussion about the techniques used in the manufacture of wall paintings in Aegean prehistory. However, most of the debate has been based on theoretical arguments, while reproduction experiments are lacking, at least in the literature. Of major importance to the subject is the work of M.A.S. Cameron, which was the inspiration for the present paper.

Cameron's work was conducted at the Department of Visual Arts of the University of Western Ontario, Canada, during the spring of 1976. However, the 'Fresco Project' was never fully published, apart from the following note in an article on the Knossian wall paintings: "...very pure lime-plasters were produced (67% lime), that were still drying out for the first time well over fifty days after painting. After final plastering, they were still malleable to some extent up to fifteen days later; they were still paintable in 'buon fresco', apparently safely, with water-mixed pigments only, up to ten days later; and they were still polishable by trowelling for at least seven days after painting" (Cameron et al. 1977, 166 n.94).

Besides the above experiment, a comparable one which took place at Pompeii in July 1911 was mentioned in the same paper. The conclusions were very similar and it was noted that further experimemation would be necessary.

Our present reproduction experiments were designed and executed by combining observations and conclusions drawn from analysis of archaeological samples (Perdikatsis et al. this volume), with the experience and the practices of painters and other scholars (Plakotaris 1987, 112-124), The results presented here should be considered as preliminary.

MATERIALS AND PROCEDURE

The extensive study of a large number of ancient samples (Perdikatsis et al. this volume) has raised several questions concerning the technique used in the plaster preparation and pigment application. As a result, we undertook a series of experimental reproductions, focussing on the time intervals between plaster preparation and pigment application within the fresco technique. A smaller scale comparison with experimental samples executed in secco technique also took place.

A number of experimental plaques were cut, and lime plaster was applied to their surface. The plaques were numbered with Roman numerals (I-VIII), whereas the different kinds of plaster were given the codes a, b and c. The three phases of the application of pigments were named A, B and C, corresponding to 1, 8 and 23 hour intervals between plastering and painting respectively. The paint layer (see detailed description below) consisted of natural pigments and laboratory prepared Egyptian Blue. Samples from the experimental plaques were examined in the same way as the ancient samples (Perdikatsis et al. this volume).

MATERIALS

The plaques on which the paintings were executed consisted of straw mixed with cement, material not very different (although lighter) from that of the original walls of the Theran houses.

Three kinds of plaster (Table 1) were used in order to verify whether the bonding of the pigments is influenced by the type of plaster lying under the final coating:

    .type a was a mortar made with slaked lime with a high proportion of calcitic sand (1:3). This mortar had already been laid onto three plaques (I, II, and V) a month before the experiment. After wetting, it received the overlying finer plasters (types b and c);

    .type b was a mixture of well-matured slaked lime and calcitic sand in ratio 1:2. It was laid onto plaques nos I, II, IV;

    .type c, the purest plaster, consisted of the same slaked lime, without any 'filler'; it was simply diluted with lime water and used as the painting surface for all plaques.

X-Ray diffraction analysis showed that the lime used for cthe plaster was portlandite [Ca(OH)₂] with a small percentage of calcite (CaCO₃).

The pigments used were all natural, except the Egyptian Blue, which was synthesised in the laboratory.

Red pigments were obtained from natural red ochres (clay with iron oxides, -hydroxides) found in the Cyclades; X-Ray diffraction analysis showed hematite, quartz, albite and alunite.

Pink was obtained by mixing a natural red pigment with dry powdered lime; minerals detected by X-Ray diffraction analysis were goethite, quartz, kaolinite and alunite.

Yellow was a commercial ochre; minerals detected are goethite, quartz, kaolinite.

Grey was produced by finely powdering glaucophane (amphibole), a natural mineral found on Syros (but not detected in Aegean wall paintings).

Black was obtained by collecting the soot of a paraffin candle.

Egyptian Blue was produced in the laboratory using the following procedure:

a) quartz (SiO₂) (63.93 wt%), calcium oxide (CaO) (14.92 wt%) in form of calcite (CaCO₃) (15 wt%) and copper oxide (CuO) (21.16 wt%) corresponding to the stoichiometrical composition of Egyptian Blue (CaCuSi₄O₁₀), plus 10 wt% lithium tetraborate (Li₂B₄O₇)as a flux, were mixed and homogenised in a mortar;

b) the mixture was fired for 20 hours at 810^oC;

c) then it was ground and re-homogenised;

d) finally, it was refired for 50 hours at 845^oC, followed by gradual cooling down to room temperature.

Egyptian Blue was the result of the following reaction:

4 SiO₂+ CaCO₃ + CuO -> CaCuSi₄O₁₀+ CO₂

The addition of 10 wt% flux, as well as the grinding and re-homogenisation resulted in total crystallisation of the mixture, producing relatively large crystals (> 200 μm) of Egyptian Blue. However, the archaeological samples contain crystals of considerably smaller size (Perdikatsis et al. this volume) and, for this reason, the synthetic material was ground to the desirable grain size (70-150 μm). This grain size produced the well known bright hue, and it was noticed that further grinding reduced the intensity of the colour, turning it to greyish.

All finely powdered natural pigments, as well as Egyptian Blue, were mixed with lime water and stored for later use.

PAINTING PROCEDURE

Plaques I and II received a layer of plaster type b, plaques III, V and VI received a layer of plaster type c, whereas plaque IV received a layer of plaster type b followed by a layer of plaster type c (Table 1).

Fifteen hours after the application of the plaster layers, the surface of all the plaques, still soft and malleable, was pressed with a trowel to give it a flat surface and a uniform consistency. The borders of the paintings were defined by impressing with a string, and preliminary sketches were drawn with yellow ochre (Fig. 4). Then these surfaces were covered with a thin, but opaque, layer of plaster type c.

One hour after the application of the plaster type c to all plaques, painting began (phase A). The paintings were executed at a room temperature of 23°C, while relative humidity was 60-70%. The other two phases of painting (B and C) took place 8 and 23 hours later respectively, under the same conditions (Table 2).

The last two plaques, VII and VIII, which were made and painted two years before the present experiments, were also examined at this stage and compared with plaques I-VI. Painting of plaque VII was executed with fresco technique, plaque VIII with secco technique and using casein as a binder. Casein was prepared from fresh soft white cheese and was mixed with lime in ratio 1:5. Plaster type a and the same earth pigments as in plaques I-VI were used in both cases.

POLISHING OF THE SURFACE

A week after the final painting, an attempt was made to polish the surface of the plaques with a pebble. However, this removed the pigments, and the plaster surface became dull. This procedure was repeated using a metallic curved object two months later, but with the same unsuccessful result. At the same time, the appearance of the final plaster layer applied to plaques III, IV,V and VI was dull when compared with the underlying plaster layer, although they both had the same consistency (type c), and especially dull when compared with the surface of the original samples. It was thought that this might be overcome through polishing of the surface, but instead it became duller. As a result, the question of the polishing of the surface still remains open.

All plaques were left to dry for a month, and then sections containing plaster and pigment were cut, polished and studied under a polarising microscope.

REMARKS

All plaques during Phase A (one hour after the application of the final coating of pure lime plaster) were not practically workable. The plaster was still too soft and affected by the action of the brush (Figs. 1a, 2a, 3a). Clearly, one hour is not enough time for the formation of an acceptable painting surface. In antiquity, this would have been to the advantage of the individuals who applied the plaster layers, since it meant that they had plenty of time before the surface was ready for the painters to start working.

Phase B (8 hours after the application of the plaster) was successful: the plaster was damp but beginning to set - that is, it provided the perfect surface for painting (Figs. 1b, 2b, 3b). It seems that the water still present in the plaster acted as a bonding agent between the substrate and the pigment, while leaving it hard enough to allow the painter to apply force on the brush without affecting the surface.

Phase C (23 hours after the application of the plaster) was not satisfactory, because the plaster was almost 'dry' (in terms of painting) and absorbed the pigments faster (Figs. 1c, 2c, 3c). This would have caused great problems for the painters because they would not have been able to draw long lines, as the pigments rapidly penetrated the surface.

Concerning the plaster-pigment bonding, the following observations are worth noting:

The pigments which bonded best to the plaster were yellow and black. This was probably due to their small grain size which resulted in a large contact surface. Red pigment came next (Figs. 2a-c) followed by pink (Figs. 1a-c), with grains which were not as finely ground as those of yellow. The pigments which bonded most loosely were glaucophane (grey) (Fig. 5) and Egyptian Blue (Figs. 3a-c). Neither of these pigments, particularly Egyptian Blue, could be ground to a finer powder without losing its colouring properties.

Pigments exhibited different levels of reaction depending on the time of application: pigments applied in phase A had penetrated significantly into the plaster because they were practically mixed with it by the brush, while pigments applied in phase B exhibited lower diffusion, and those applied during phase C showed little or no fusion at all. This last observation was most pronounced for the blue-grey pigments which had the largest grain size.

Examination of the polished sections of the red and yellow pigments of plaque VII (fresco technique) revealed that yellow ochre (Fig. 6a) had penetrated deeper than red ochre (Fig. 6b), again due to its smaller grain size. As far as plaque VIII (secco technique) is concerned, the paint layer was thick but no pigment diffusion was observed (Fig. 7).

CONCLUSIONS

The observations of the above experiments are presented in Table 3 and can be summarised as follows:

   1) On the basis of the results obtained, it appears that painting is not influenced by the nature of the plaster layers existing under the final coat.

   2) It is not possible to paint over a freshly applied layer of plaster, because the wet plaster is removed by the brush and mixed with the pigments, which consequently lose their bright hues.

   3) In the conditions under which our experiment was conducted, it was observed that the ideal waiting time between plaster application and painting was about two hours. After this interval, it was possible to work with ease for 6 to 8 hours. After a maximum period of 10 hours, the plaster became too 'dry' to bond adequately with the pigments, especially the blue one, some grains of which could be removed when rubbed with wet cotton. The rest of the pigments exhibited strong bonding compared to the Egyptian Blue, although the penetration was not high.

The main factors that seem to influence the application of the pigments on the plaster are:

a) moisture of the plaster; b) grain-size of the pigment. It is apparent that the finer the pigment the better the penetration achieved.

The macroscopic and microscopic appearance of the plaques, when compared with the original ancient samples, show some remarkable similarities in the way the pigments bond with the plaster. As has been shown, for every successful sample from the reproduction experiments a parallel among the ancient ones was found.

----------------------------------------------------

------------------------------------------------