Chromatographic analysesThe paper focuses on fatty acids and biomarkers that can directly indicate the source of the examined organic residues. The identified fatty acid contents in samples from the Domasław cemetery and the Milejowice settlement are included in Table 2.
Table 2 Fatty acids determined in the studied samples (content in %); (-) under the limit of quantification.Eighteen acids, primarily saturated, were determined in the samples. C16:1 and C18:1 represented the group of unsaturated acids. C18:2 was determined in sample 113 from Domasław. C6:0, C8:0, C9:0, C10:0, C12:0, C16:0, C18:1 and C18:0 acids were present in all samples. Long-chain acids—C20:0 and/or C22:0—were detected in most samples from Milejowice, and in six from Domasław. C11:0 acid was identified in vessels 42 and 43 from votive pits in Milejowice. This acid was also rare at the cemetery, recorded in urns, rhyta, a censer, and offering vessels. Statistical analyses showed that settlement and burial vessels differentiate most acids C13:0, C15:0, C22:0 and the C22:0/(C16:0 + C20:0) ratio.Any conclusions about the sources of the examined organic residues based solely on the presence and content of individual acids would be unreliable. Most acids are present in both plant and animal resources. Moreover, the residue composition can change over time, which should always be considered in archaeological samples. Consequently, researchers are looking for another way to interpret the results of chromatographic analyses. One of the methods is based on studying the proportions of selected fatty acids, as it appears that acid proportions can remain unchanged over time. The conclusions presented in this paper were based on the work of Eerkens20. Based on his proposed proportions, the most likely sources of the studied residues were designated. The calculated proportions of selected acids are shown in Table 3.
Table 3 Calculated proportions of selected fatty acids in the studied samples.The acid proportions suggest that most examined samples have a mixed origin. The residue probably came mainly from seeds, nuts, and berries in combination with fat from land mammals (samples 39, 42, 47, 55, 58, 61, 79, 86 from Milejowice and 3, 4, 12, 91, 95, 104, 105 from Domasław). The proportions in grave samples 90, 93, 94 suggest seeds, nuts and land mammal fat. Vessels 40, 43, 46, 49, 53, 72 from the settlement and 10, 101, 106, 108, 111, 113 from the cemetery were of plant origin (nuts and seeds, berries), 70 from Milejowice – originated from seeds and nuts. The C16:0:C18:0 ratio has long been used to distinguish different plant oil sources, with a high (> 3) ratio considered characteristic of poppy seed oil21. At vase 105, the ratio was 6.98. A higher C18:1 ratio was detected in rhyta, vases and pots from the cemetery, including most vessels from grave No. 7429 with vases 101 and 104. Scholars suggest that dicarboxylic acid compounds are more likely to be found in plant oils22. The presence of dicarboxylic acids and oleic acid (C18:1) increases the possibility that the samples contained acid-rich plant oil or derivative mixtures.Gas chromatography combined with mass spectrometry made it possible to determine organic compounds from different groups. Based on a literature review, it was possible to select compounds that can be considered archaeological biomarkers. Biomarkers are associated with one specific material or substance, regardless of its origin—plant, animal or mixed. The detected fat is not necessarily an oil but may have been an ingredient of herbs, vegetables or grains.Methyl dehydroabietate, a dehydroabietic acid derivative, was detected in almost all samples from Milejowice (except 42, 43, 55) and seven from Domasław (3, 4, 12, 90, 91, 93, 94). This acid is classified as a resinous acid and is considered an indicator of resin or its products23,24,25. Resins have hydrophobic properties and can be used to seal unglazed vessels26, but they have also been intentionally added to alcoholic beverages to preserve, enhance, and change their flavour27,28,29,30. Adding resin to wine to protect it against disease, for medicinal purposes and to cover up off-tastes and off-aromas was a popular and widespread practice throughout the ancient world31. Cedrol—sesquiterpene alcohol, found in the essential oil of coniferous trees, was present in settlement vase 55.Glycerol, a product of lipid degradation, was also observed in most samples from Milejowice (except 55, 79, 86) and in more than half of the samples from Domasław (3, 90, 91, 93–95, 101, 104, 106). Vanillin, a phenolic compound, and vanillic acid (found in elderberry juice, blueberries, strawberries, and alcoholic beverages32) were detected in nine vessels from Milejowice (39, 40, 46, 53, 55, 58, 61, 70, 72) and in all samples from Domasław except 101, 106, 108, 111, 113. Vanillin may originate from tree resin or pine wood/dust, which is documented in the same samples by the presence of methyl dehydroabietate. Acetowanillone, identified in settlement vase 70, a structural analogue of vanillin, is found mainly in wines aged in oak barrels.Lactic acid, determined in nine vessels from the settlement (39, 40, 42, 46, 47, 49, 58, 61, 70) and eight from the cemetery (4, 10, 90, 91, 93, 94, 95, 104), is one of the essential products of bacterial fermentation, which occurs, e.g., during vegetable pickling or milk fermentation. However, malic acid (a beer and wine ingredient) can also be converted by bacteria into lactic acid, which gives the beer a sour taste33,34.Azelaic acid was identified in nine vessels from the settlement (39, 40, 42, 49, 53, 55, 58, 61, 86) and seven from the cemetery (4, 10, 12, 93, 94, 101, 105); oxalic acid in four from Milejowice (40, 46, 55, 58) and three from Domasław (12, 91, 95); suberic acid in five from the settlement (42, 53, 55, 58, 61) and two from the cemetery (4 and 12). Their presence may suggest that the studied residues originated from grain products, including wheat, rye, or barley35,36.Suberic acid is produced also during the oxidation of castor oil and is used to produce resins. Acetic acid, observed in vase 4, may indicate the contact with fermented food34. It is formed from ethanol under the action of aerobic acetic bacteria (vinegar production), as well as from acetaldehyde, which, as a component of beer, gives it its characteristic flavour and aroma33,34,35. Fumaric acid, confirmed during beer fermentation and wort production34, was present exclusively in vase 94 from Domasław. Only settlement vessel 53 produced traces of caprolactone, a fragrance component found in flowers, some fruits and vegetables. Δ-Caprolactone is found in heated milk fat37.Stigmastanol, present in four samples from Domasław (4, 12, 93, 95) and seven from Milejowice (40, 43, 55, 58, 70, 79, 86), also indicated their plant origin. This sterol occurs in vegetable fats or oils of many plants: beans, rapeseeds and herbs. Another noteworthy example is benzoic acid, determined in half of the samples from the necropolis (90, 91, 93, 94, 101, 104, 106, 113) and only one from the settlement (53). Its presence may also indicate that the examined residues originated from plants, as it is a component of plant tissues, especially fruits and vegetables38. This acid is detected, for example, in cherry bark, raspberries, and honey. Benzoic acid may also be a degradation product of anthocyanins by ketones in wine39.Of particular importance may be the tiglic and croton acids present exclusively in vase 12 from the Domasław cemetery and 13 containers from this site, such as rhyta, a kernos, censers, a disc-plate, an urn, and offering vessels. Tiglic acid is a compound with a spicy smell, it occurs, among other things, in croton oil obtained from the seeds of laxative croton, a plant from the Euphorbiaceae family, used for medicinal purposes.Borneol, which appeared in two vases from the cemetery (10, 90) and four from the settlement (39, 42, 55, 79), is a compound from the terpene group with a camphor-like fragrance. It is a component of many essential oils, pine resin, and herbs, e.g., tansy, savoury, sage, and plants from the wormwood family. Like croton oil, it has medicinal and toxic effects and may cause eye and skin irritation and vomiting. Borneol was found mainly in the most distinctive vessels from the cemetery, such as an urn, a rhyton and offering vessels, At the Milejowice settlement, it occurred in dippers, a plate-disc and bowls. Phenoxyethanol was detected only in funerary vases 90, 91, 93, 94 and 95, two miniature rhyta and an urn. It can be found naturally in some plants, such as onions and chicory, and helps control and prevent the growth of bacteria, yeasts and moulds. Inulin, an extract from chicory root, has been used as a sweetener. Some beer brewers use roasted chicory to add flavor to stouts, or to augment the hops. Roots contain essential oils similar to those found in plants in the related genus Tanacetum. Carvacrol was detected only in three vases from Milejowice (55, 58, 61). There are many sources of carvacrol, including various kinds of thyme and marjoram.Adipic acid from three funerary samples (90, 91, 101) occurs naturally in beets and sugarcane. Gramine (donaxin), found in funerary vase 101, is a naturally occurring indole alkaloid found in several plant species, such as silver maple and Hordeum, a genus of grass that includes barley40. 9,12-Dihydroxyoctadecanoic acid (Z,Z), present in a funerary vase with a funnel-shaped rim and conical neck (101) and in all vases with cone-shaped necks (105, 106, 108), is a linoleic acid derivative and a component of most vegetable oils and animal fats, found in many foods, e.g., legumes, cloves and nuts. It is produced through the hydration of ricinoleic acid, which is found in large amounts in castor oil and has historically been reported to be a significant resource with various uses in ancient times.3,4-dihydroxybutyric acid found in vase 72 from Milejowice, is believed to be formed via the degradation (cooking) of di- and polysaccharides, including lactose. Farnesol, an alcohol from the terpene group obtained from linden flower oils, is used mainly as the lily-of-the-valley fragrance and is a component of many essential oils, including lily of the valley, linden and acacia. It was found in painted vases 47 and 49 from the settlement.The contents of funerary vessel 113 proved very distinctive, suggesting the presence of plants and honey. The detected spirost-8-en-11-one, 3-hydroxy- is a component of sugar plants. The sample also contained 7,8-epoxylanostan-11-ol, 3-acetoxy, which is a cholesterol-based alcohol with anti-inflammatory and antibacterial properties that occurs in many plants. Furthermore, a propolis ingredient, 17-pentatriacontene, 1-heptatriacotanol, was identified. Additionally, 1-heptatriacotanol, an alcohol with antioxidant, anticancer, and anti-inflammatory properties, was confirmed. It occurs in various waxes, e.g., beeswax or plant waxes (jojoba oil). Last, the studies detected 1-monolinoleoylglycerol—an ether, glyceride consisting of one chain of fatty acid (linoleic acid) and glycerol. It occurs, among others, in plants of the Lamiaceae family, which are used as spices, medicines and teas.Valproic or valeric acid was present in vases 90 and 91 from the cemetery and vessels 42 and 43 from settlement votive pits. Valproic acid is a saturated monocarboxylic acid and a valeric acid derivative. It occurs naturally in some foods, such as berries. It has sedative and anticonvulsant properties.The presence of dibutyl phthalate, a phthalic acid present in six vessels from Milejowice and nine from Domasław, may indicate the thermal processing of products in vessels41 or may result from sample contamination.Acid data analysis and classificationIn this study, based on acid-related data, we applied the FCM algorithm to cluster samples of ceramic vases from the settlement in Milejowice and the cemetery in Domasław. Features important for the data analysis were selected by t tests with a threshold of 0.05 and VIP values were obtained with PLS regression (Fig. 2).Figure 2Important features selected for sample classification: (a) features of subset 1 selected by t-tests with a significance threshold of 0.05. Blue circles represent features above the threshold (p); values are transformed by –log10; (b) Features of subset 2 selected by the VIP value (score above 1). Characteristics of the biomarker distribution in vase samples, including a grouped histogram of biomarker occurrences in samples from the cemetery and settlement (c) and a clustered heatmap of biomarkers for which the minimum absolute difference between the two groups of vases in histogram was at least 2. Additionally, the frequently occurring lactic acid was considered (d).The results were two subsets of variables used independently in the analysis. Subset 1 includes acid-related variables such as C13:0, C15:0, C22:0 and C22:0/(C16:0 + C20:0) (Fig. 2a,b). The set of variables with the highest VIP value (> 1) included the same components, except for the additional C12:0/C14:0 acid proportion. The test values are listed in Table 4.
Table 4 Values of the t-statistics of the Student’s t-test for raw data regarding acids and their proportions in two groups of vases from different locations (independent groups).Using the fuzzy C-means method, the dataset was grouped into two clusters, where each data point belonged to each cluster to some extent. Clustering results are shown in Fig. 3a and Fig. 3b. The obtained results show that the clusters largely correspond to the sample origin. The probability plot for feature subset 1 (Fig. 3c) shows that cluster 1 contains samples from the necropolis and five samples from the settlement, while cluster 2 contains only samples from the settlement. Adding acid proportion C12:0/C14:0, another considered feature slightly reduced the probability that the samples belonged to the selected clusters (Fig. 3d) but did not significantly disturb their previous distribution. With the data in Fig. 3c,d, adopting more flexible grouping rules, e.g., more restrictive than a classification cut-off at a probability of 0.5 was possible. The requirement of a higher probability of class membership leaves some samples out as unclassified due to uncertainty.Figure 3Illustration of the clusters of archaeological samples from the cemetery and the settlement in the plane of the PCA main components for the selected feature subset 1 (a) and subset 2 (b). Each ellipse is a 95% confidence ellipse for a 2D normally distributed data set. The probability of samples belonging to clusters 1 or 2 based on the selected acid data is also presented for feature subset 1 (c) and subset 2 (d). Confusion matrices for feature subset 1 (e) and subset 2 (f).Figure 3e,f represent the confusion matrices obtained for the two feature subsets. They confirm the obtained probability results and quantify the quality of the archaeological sample classification. The ratio of the sum of diagonal elements to the sum of all matrix elements represents the classification accuracy obtained from clustering. For both matrices, it is 0.84 and 0.81, respectively. It means that features selected by the Student’s t-test provide a better separation of clusters than features selected by the PLS-DA method and, simultaneously, ensure better compliance with their actual division.Table 5 includes three different indicators of fuzzy clustering effectiveness explained in Eqs. (1)–(3).
Table 5 Evaluation of the clustering results for selected acid data.Dunn’s partition coefficient (DC) reaches 0.79, which suggests that the clustering is not very fuzzy, and the crisp sample grouping methods can also be applied. Partition entropy index PEI = 0.35 in the range [0,1] informs that the uncertainty within the proposed partition is relatively low, which confirms the observations about the DC. The Fuzzy Silhouette Index value of 0.76 leads to the same conclusions.Biomarker data clusteringThe vessels yielded different biomarkers, most notably benzoic acid, vanillin, and phenoxyethanol. The analysis of histogram and clustering results revealed that some biomarkers were more common in settlement samples (Fig. 2c,d). These included methyl dehydroabietate, glycerol, azelaic, suberic, adipic acids, and stigmastanol. Vanillin, and benzoic acid were most frequent in cemetery samples. Grave vases had a higher plant oil content.Vases were most differentiated by the benzoic acid, which were detected in 62.5% of funerary vessels and only 6.5% of settlement vessels. Lactic acid was present in equal proportions in settlement vases and funerary vases with funnel-shaped rims. However, there was no lactic acid or dibutyl phthalate in grave vases with distinguished conical necks. In contrast, 9,12-octadecadienoic acid occurred in all vases with conical necks, one vessel with a funnel-shaped rim, and in none of the settlement containers.The cluster analysis of the identified biomarkers showed that vessels from the settlement were more similar to one another than were vases from the cemetery. Sample 113 differs from the others and is closer to the pattern recorded in settlement vases.
Funerary vs. domestic vessels from the Hallstatt period. A study on ceramic vases from the Milejowice settlement and the Domasław cemetery
