What do we know of Roman wall painting technique? Potential confounding factors in ancient paint media analysis

During recent decades, the theory that fresco was the most common painting technique in Roman murals has enjoyed general acceptance among specialists [1–3]. As a result, it is now common practice in museums to state that fresco was the technique used in Roman murals within their collections [4–10]. However, recent studies of the paint media used in these paintings seem to question the validity of the fresco theory, as most current analyses detect the presence of organic binders. These binders are as varied as wheatpaste, egg and emulsified beeswax. These studies have opened a new perspective on Roman wall panting techniques, which is the subject of the present work.

The diversity of paint media identified seems to suggest the use of various painting techniques in Roman murals. According to this hypothesis, it would be expected that, when studies analyse samples from different archaeological areas and periods, various types of binders will be identified. However, contrary to those expectations, the results of such studies have proven rather homogeneous.

Corso, Gelzo and co-workers analysed the composition of binders in unrestored Roman wall paintings from Villa Imperiale, Pompeii (first century AD) [11], from Liternum, Italy (first century AD) [12], and in fragments of the four Pompeian styles (200 BC to 79 AD) from the Marcus Fabius Rufus House, Pompeii [13, 14]. The results indicate the use of a wheatpaste-based paint in all the samples that allowed for the identification of the binder. These studies are probably the only ones detecting this technique Roman wall painting. Yet remarkably, this uncommon technique was found in most paintings studied by this research group, raising questions about possible causes behind the singularity of these results.

A research team with which the author has collaborated analysed paint binders in unrestored Roman wall paintings from Ampurias, Cartagena, and Baelo Claudia, Spain, dating from the first century BC to the second century AD [15]. In all the samples studied the paint medium was hypothesized to be emulsified beeswax. This study of paint samples randomly selected from different archaeological areas seems to point to extensive use in Roman murals of a binder made from emulsified wax. This extensive use seems to be confirmed by similar results obtained in a study carried out by the same group on unrestored Roman wall paintings from Mérida, Complutum and Marsala which will be discussed later. As in the case noted in the previous paragraph, these results are inconsistent with those found in the literature, as the presence of emulsified wax has not been reported in other studies.

Another research group has further analysed the binders in wall paintings at the Villa of the Papyri [16], and in a painted wooden ceiling at the House of Telephus Relief, both in Herculaneum (and dating to before 79 AC) [17]. In both paintings, results indicated the use of egg tempera. This group has also analysed Mycenaean wall paintings at the Palast of Nestor in Pylos, Greece (before 1180 BC) [18], on an Etrurian wall painting (fourth century BC) [19], and on an Etrurian sarcophagus of painted alabaster (fourth century BC) [20]. Again, egg tempera was hypothesized, suggesting widespread use of this technique in ancient Mediterranean cultures, both in wall painting and on other types of supports. It is significant that the studies mentioned in the two preceding paragraphs have not detected the presence of egg in any of the samples studied, and that some analysis specifically reports lack of egg in the murals of Pompeii and Herculaneum [21]. As in the two previous cases, the results obtained by these researchers were consistent with each other, but inconsistent with findings by other teams.

Elsewere, Casoli and co-workers analysed the organic binders in wall paintings from various houses in Herculaneum [22] and Pompeii [3], and the painting technique was hypothesized as being fresco in all the samples. This technique has not often been identified in studies analysing organic binders in Roman wall painting. Therefore, the ratio of samples ascribed to fresco in these studies seems to be remarkably high compared to other studies, This high ratio between samples ascribed to a specific painting technique by a particular research group is a bewildering characteristic common to most studies of Roman painting techniques, again rising questions about possible causes.

Considering the different paint media identified in Roman wall painting, the similarity in the binders detected by each team in samples of various wall paintings is worth mentioning. This similarity may suggest the existence of a common painting technique, characterized differently by each research group. The hypothesis of the existence of a dominant painting technique serves as a theoretical background for the current theory of fresco, based on the common technical characteristics displayed by Roman wall paintings.

The clustering of each type of binder around a specific research group, shown in Table 1, appears highly unlikely. It may indicate the presence of confounding factors in the analytical approach used, thus causing skewed results.

^aResearch teams sharing members

The possible presence of these factors is further emphasized when different researchers analysing the same samples come to different conclusions. One relevant example can be found in the Roman wall paintings from Casa del Mitreo in Mérida, ascribed to emulsified beeswax paint by one research team and to fresco by another [23, 24].

Which potential confounding factors could be impacting the puzzling results obtained in analyses of Roma wall paint media? Most studies of ancient binders acknowledge considerable difficulties in accurate identification due to the degradation of the original compounds, the mixture and interaction of materials present in the preparation, paint and protective layers, and the small sample size [25–27]. Besides these factors, often taken into account in such studies [28–34], are two other factors, not usually considered in literature, which may have a decisive influence on results: the difficulty in efficiently extracting organic binders from ancient paint samples, and the presence of organic material due to microbiological contamination.

The first time the author of this work carried out a study to determine the composition of binders in Roman murals, the extraction method used was developed in the same laboratory in which the study was performed¹, and it was successfully used to analyse the composition of paints dating from recent centuries, as had been commonly studied in that laboratory [35]. The Roman samples analysed were taken from unrestored Roman wall paintings from Mérida and Complutum (in Spain) and Marsala (in Sicily). The analysis of these samples by GC–MS showed no presence of organic material, suggesting that an inorganic paint medium had been used. This data together with the identification of calcium carbonate in the paint layers seemed to indicate that the Roman murals under analysis were painted with fresco. In order to check for presence of organic compounds in the samples, the inorganic residue of the extractions was subsequently analysed by FTIR. The spectrograms indicated presence of organic materials, suggesting that the extraction method was not appropriate for ancient paint binders. After subsequent use of a more efficient extraction method, analysis by GC/MS allowed the hypothesis that emulsified beeswax could have been the paint medium used in all the murals analysed [23].

The difficulty in extracting ancient paint media and its possible influence in the analytical results suggests that this issue should be taken into full account when characterizing ancient painting techniques. However, the absence in most of the studies of remarks on the use of extraction methods suitable for ancient binders, or of tests to assess their efficiency, may generate doubts about the suitability of the extraction methods used in some analysis.

Table 1 shows that some studies are carried out with extraction yields of less than 0.1% of the paint weight. This amount of binder would seem insufficient to fix the pigment to the support, or to explain the stability of the painting, thus allowing the hypothesis that part of the binder may not have been extracted, which would produce an incomplete characterization of its composition. This possibility might further suggest a link between the extraction methods used and the results obtained, making advisable an assessment of the efficiency of the extraction methods used in characterizing ancient paint media.

Moreover, the usual presence of microorganisms in ancient wall paintings can provide an additional source of organic material which could be mistaken as part of the original binder. This presence, not generally considered in analyses of Roman wall painting, may modify the results of an analysis, particularly when combined with low extraction yields.

Studies of ancient organic binders which identify the painting technique as fresco are of particular interest. This painting technique does not require any organic binder, so it would seem reasonable to ascribe the lack of detection of such binders to a painting executed in fresco. However, if the extraction method used was not efficient enough to detect the presence of an ancient organic binder, the conclusions reached could be inaccurate.

In order to assess this possibility, it is useful to review the paint analysis in Greco-Roman polychrome sculpture. As the paint used in sculpture requires an organic binder to fix the pigments to the support, the lack of identification of organic binders has to be ascribed to the use of a fresco painting technique, and may instead be taken as proof of the difficulty to characterize ancient paint media. A review of the literature on ancient polychrome sculpture shows that, often, studies analyse only pigments [36–39] or are unable to identify the binder [40–42].² The result is a lack of understanding of the painting techniques used in Greco-Roman polychrome sculpture, as summarized by Kenneth Lapatin: “ancient binders have yet to be identified, and the original painting and finishing techniques are far from well understood” [43]. Transferring these results to mural painting, it could be hypothesized that the reason some chemical analyses of Roman wall paintings have not detected any organic binder is due not to execution in fresco, but to the difficulty in extracting the original paint medium.

Also noteworthy are studies of Roman wall paintings in which organic compounds are detected in some samples but not in others found in excavation. As the compounds identified in these studies are usually located on the outermost layer of the samples, it seems reasonable to hypothesise that the original technique was fresco, and that the presence of organic material is due to modern conservation treatments. This is the case in the aforementioned studies by Casoli and colleagues. The detection in these analyses of organic compounds such as wax, egg or animal glue seems to confirm the use of a suitable extraction method. However, if the method used was not suitable to extraction of ancient binders, subsequent analysis might detect the presence of modern binders although they were unable to detect the presence of ancient paint media. Therefore, in this case, it would be necessary to determine the efficiency of the extraction method used.

The most critical consequence of the lack of data on the efficiency of the extraction methods of Roman paint binders, and the possible influence of microbial contamination, is the difficulty in knowing at this point whether some results are inaccurate, and to what extent, fomenting an overall uncertainty on Roman wall painting techniques. Such uncertainty can impact not only upon our knowledge of Roman art and technology, and the efficient use of resources in paint media studies, but also the conservation of the paintings, since the “knowledge of the original painting techniques [… is] essential in the implementation of appropriate conservation projects” [44–46].³⁴