# Analysis of medium-sized pouring vessels

### Introduction

This article investigates artefact variability of Attic Geometric pouring vessels. Macroscopic analysis is conducted on 103 ceramic artefacts in total. These are 84 vessels with complete profiles and 19 incomplete vessels or sherds. The majority of this material (73 artefacts or 70.9%) comes from Athens, while 30 pieces (or 29.1%) have been identified as broadly Attic by Coldstream (2003b; 2010).

### The argument

This article argues that according to metrical features and proportions, the production of pouring vessels became highly standardised only after the beginning of the Late Geometric period. During that time, new shapes such as neck-less trefoil oinochoai and pitchers appeared for the first time next to standard trefoil oinochoai with necks. Still, all types of pouring vessels were the products of the same chaîne opératoire and followed similar conceptualisations. Late Geometric standardisation is likely to suggest a reduction in the numbers of ceramic workshops producing pouring vessels, or a conscious shift towards specialisation in the production of such shapes as a result of increasing consumption demands. The most standardised feature of all pouring vessels is their fabric: all pots have been produced from the same clay, resulting in the same two variants noted in the case of large closed ceramic containers. The use of a single fabric remained unchanged for at least two centuries. The external treatment of elaborately decorated containers moved towards gradual abandonment of thick lustrous coating after MGII-LGIa, which coincided with the generalised use of figurative decoration. Despite this technological change in external treatments, all pouring vessels demonstrate similarities, pointing to a chaîne opératoire that was controlled by few and highly specialised potters or workshops. Such production units followed strict technological traditions and they probably clustered together in a single production site, particularly after LGIa.

### Analysis of metrical features and proportions

*The Athenian Agora (supplemented by the Kynosarges burials and the collections of the British School at Athens)*

For the analysis of metrical features (metrics) and proportions of pouring vessels, 37 artefacts with complete profiles were selected and analysed. In this assemblage, 29 vessels come from the Athenian Agora, 1 from the Kynosarges burials and 7 from the collections of the British School at Athens. Pottery from Kynosarges and the British School’s collections is examined together with the Agora artefacts due to their broader decorative and stylistic similarities. With regard to its typological variation, the assemblage consists of 23 decorated trefoil oinochoai (9 of which are neck-less) and 6 decorated pitchers, all recorded in Charts 1 and 3. Only 4 vessels with complete profiles come from burial contexts; 27 come from mixed non-burial deposits, while the archaeological context of 7 vessels is unknown. In addition to this study, Chart 2 presents another 17 pieces coming from incomplete or fragmented pottery from the Athenian Agora, including 1 sherd from Kynosarges. These fragments are not used in the analysis of metrical features and proportions; however, they supplement the analyses of fabrics and decorative technology further below.

In the above ceramic assemblage, the typical comparison between rim diameters and net heights that took place for large containers in is not applicable. This is because rim diameters cannot be recorded on trefoil oinochoai due to the recessed shape of their lips (or trefoil mouths); still, rim diameters are recorded for pitchers (Chart 1). The only direct comparison between the two typological groups relates to base diameters and net heights, as these features are recorded on both shapes.

*Figure 1: Scatter-graph of correlation between net height and base diameter. Athenian/Attic pouring vessels from the Agora, Kynosarges, and the British School’s at Athens collections with complete profiles.*

Figure 1 presents the correlation between base diameter and net height for the above assemblage. The scatter-graph shows that trefoil oinochoai exhibit greater variability compared to pitchers, which appear standardised and distinct. All vessels form five clusters with specific typological and chronological properties, and the borderline between oinochoai and pitchers is at 33 cm net height.

Cluster 1 consists of 5 pots with different shapes: standard, broad and neck-less broad trefoil oinochoai. It is the only cluster containing broad oinochoai, which were produced in small heights with narrow bases. No distinct chronological pattern is noted in this group:

Cluster 2 is the densest of all clusters and shows two distinct properties: Firstly, the majority of vessels (12 out of 15) date in the Late Geometric period; secondly, it is the only cluster containing neck-less trefoil oinochoai:

Cluster 3 consists of 6 pots coming from the period between EGII and MGI. This group of standard trefoil oinochoai contains the tallest and widest vessels in the entire assemblage. Vessels properties resemble those of the cluster with the largest decorated neck-handled amphorae from Kerameikos and both clusters date in the same period:

Cluster 4 consists of 4 vessels that have narrow bases yet they are relatively tall. According to the shapes and chronological distribution of this group, no distinct pattern can be noted:

Cluster 5 contains every pitcher in the entire assemblage. Such vessels have standard base diameters and their heights exceed 33 cm. They all date in LGII:

Finally, Oinochoe P6401 from the Athenian Agora stands out and must be treated as a loner. This vessel is of average net height; however, it has the narrowest base diameter in the entire assemblage.

The presence of distinct groups of pouring vessels, which not only cluster according to their typology but also according to broader chronological periods, shows that any further analysis should accommodate two requirements: firstly, Late Geometric vessels must be examined separately; secondly, neck-less oinochoai must be treated as a distinct typological class. The analysis of mean proportions across two broader chronological groups based on the data of Chart 3 shows fluctuations in standard deviations. More specifically, the proportions of neck length to net height and base diameter to net height for oinochoai produced after LGIa are smaller compared to those for oinochoai produced before LGIa. The opposite pattern is noted with regards to the proportion of handle attachment height to net height; however, it must be noted that the Late Geometric mean includes different typological sub-classes. As it is explained further below, once these typological subgroups are separated, Late Geometric pots show homogeneity with regard to all their proportional features:

Furthermore, most mean proportions of Late Geometric pitchers show lower standard deviations compared to those of trefoil oinochoai from the same period:

According to the comparison of standard deviations, trefoil oinochoai between EGII and MGII-LGIa exhibit greater artefact variability compared to those produced after LGIa. Furthermore, the clustering of LG oinochoai in Figure 1 suggests larger degree of standardisation in their production compared to their earlier equivalents. Finally, the most standardised of all pouring vessels are pitchers.

The possibility of standardisation in the production of Late Geometric pouring vessels is also noted with regard to base shapes. In a total assemblage of 39 oinochoai with recorded base diameters from three sites (31 presented in Chart 1 and 8 in Chart 2), 17 have been produced with ring bases. An example of ring base (or ring foot) is presented in Figure 2.

*Figure 2: Example of ring base on Early Geometric I oinochoe P3687 (Papadopoulos 2003, 100).*

During the chaîne opératoire steps for the production of oinochoai (Figure 3), the ring base was formed during a separate episode on the potter’s wheel, which required the vessel to be inverted; therefore, forming such shapes without ring bases reduced the time of production by cutting down one of the chaîne opératoire steps. In the present assemblage, the majority of oinochoai with ring bases come from the period between EGI and MGII (14 out of 17):

This suggests that during the Late Geometric period production moved towards greater simplification of the chaîne opératoire and ring bases declined.

*Figure 3: The assembling process of Geometric trefoil oinochoai.*

A similar move towards simplification during Late Geometric times is noted with regard to the increase of neck-less shapes. Chart 1 shows that more than half of the Late Geometric trefoil oinochoai with complete profiles (9 out of 17) are neck-less. According to Figure 3, the manufacture and attachment of necks on oinochoai took place during two separate episodes on the wheel’s head; therefore, the production on neck-less vessels reduced the total chaîne opératoire by another two steps.

*Figure 4: four types of oinochoai examined in this article.*

According to the comparison of the four oinochoai shapes examined in this thesis (Figure 4), vessels with necks and ring bases are the tallest and most complex of all. By contrast, broad or neck-less vessels with flat bases are shorter and simpler in their conceptualisation and execution (sensu Van der Leeuw 1994, 136-7); therefore, their production probably required less time and effort. As neck-less flat-based pouring vessels increased after LGIa, it is likely that the preference in simpler shapes and the reduction of the chaîne opératoire steps were dictated by increasing consumption demands. The same demands were probably responsible for greater standardisation in the production of pouring vessels.

The low standard deviations for the proportions of handle attachment height to net height (6.69) and neck length to net height (4.13) for oinochoai after LGIa are likely to suggest technological traditions followed in the production of such finewares. The same hypothesis is also likely for pitchers, where standard deviations for most mean proportions range at small percentages, with exception of base diameter to rim diameter. Both cases require further investigation.

*Figure 5: Scatter-graph of correlation between net height and neck length. Athenian/Attic pouring vessels from the Agora, Kynosarges, and the British School’s at Athens collections with complete profiles.*

Figure 5 presents the correlation of neck length to net height for 24 pouring vessels with necks and complete profiles. Nine neck-less trefoil oinochoai have been left out. According to the graph, the regression lines for oinochoai dating before and after LGIa appear close together. The regression line for the group before LGIa follows the equation y = 0.3565x + 0.3913 (where y = neck length and x = net height). In other words, the neck lengths of these oinochoai are roughly equal to 35.65% of a vessel’s net height. The difference of 0.3913 cm is too small to be considered. The coefficient of determination of the regression line (R2=0.7171) shows weak statistical correlation (71.71%).

A similar pattern is noted for the regression line of trefoil oinochoai produced after LGIa. There, the regression line follows the equation y = 0.3183x + 0.6431 (where y = neck length and x = net height). In this group, neck lengths are roughly equal to 31.83% of a vessel’s net height, while the difference of 0.6431 cm is too small to be considered. Similarly to the regression line for oinochoai before LGIa, the coefficient of determination (R2=0.7218) shows weak statistical correlation (72.18%).

The proportional pattern for LGII pitchers is unclear. The regression line for 6 vessels in Figure 5 follows the equation y = 0.4215x – 1.2491 (where y = neck length and x = net height). In other words, neck lengths of Attic pitchers are roughly equal to 42.15% of their net height reduced by a difference of 1.25 cm. The coefficient of determination of this regression line shows 72.23% statistical correlation (R2=0.7223), which is equally weak to the correlation of trefoil oinochoai.

According to the above comparison, it is likely that similar patterns might have existed in the conceptualisation of necks between oinochoai and pitchers; however, these cannot be securely confirmed due to the ambivalent statistical correlations of the regression lines.

*Figure 6: The assembling process of Geometric pitchers.*

From a chaîne opératoire perspective, both pouring vessels follow a similar sequence in their assembling processes; therefore, they share common features of partonomy. The assembling process of oinochoai followed at least six steps before LGIa, or at least three steps after LGIa, when ring bases and necks were not popular any more (Figure 3). The same assembling process for pitchers included at least five steps, despite the fact that such vessels did not have ring bases and trefoil mouths (Figure 6). The necks of both typological classes were formed and attached during separate episodes on the potter’s wheel; therefore, it could be likely that potters shared some similar conceptualisations with regard to neck lengths. In the above assemblage these conceptualisations cannot be clearly mapped and more samples are required. The situation is different with regard to handle attachment heights.

*Figure 7: Scatter-graph of correlation between net height and handle attachment height. Athenian/Attic pouring vessels from the Agora, Kynosarges, and the British School’s at Athens collections with complete profiles.*

Figure 7 presents the correlation of handle attachment height to net height for 36 out of 37 pouring vessels with complete profiles from the three sites. Neck-less trefoil oinochoe P12115 is partly reconstructed with plaster and its handle attachment height could be wrong; therefore, it was left out. Figure 7 also plots the correlation for neck-less oinochoai separately due to their distinct shape. According to the graph, the regression lines for the two oinochoai groups dating before and after LGIa follow two different patterns. The regression line for the standard oinochoai group before LGIa follows the equation y = 0.6044x – 1.0072 (where y = handle attachment height and x = net height). The handles of these oinochoai are attached roughly at 60.44% of a vessel’s net height with a difference of 1.0072 cm. The coefficient of determination of the regression line (R2=0.8432) shows 84.32% correlation of variables, which is relatively satisfactory.

The regression line for standard trefoil oinochoai produced after LGIa suggests a different pattern: the line follows the equation y = 0.6565x – 1.8005 (where y = handle attachment height and x = net height). In this group handles are attached at roughly 65.65% of a vessel’s net height reduced by an average of 1.8 cm. By contrast to oinochoai produced before LGIa, the coefficient of determination of those dating after LGIa (R2=0.9793) shows 97.93% statistical correlation, which is nearly perfect. The regression line for neck-less trefoil oinochoai produced after LGIa suggests no clear pattern. It follows the equation y = 0.7408x – 2.4625 and the coefficient of determination (R2=0.6308) shows weak statistical correlation.

According to the above, it is likely that the handles of standard trefoil oinochoai were attached with some specific conceptions that could relate to distinct technological traditions: before LGIa handles were attached at roughly 60% (or 4/5) or a vessel’s net height, while after LGIa at roughly 66% of a vessel’s net height. The second pattern shows similarity with the proportions of handle attachment height to net height of neck-handled amphorae, noted in Chapter 4: both wares have their handles attached at roughly 2/3 of a vessel’s net height (between 65% and 68%). The coefficients of determination suggest that the group of standard trefoil oinochoai after LGIa shows less variability compared to that before LGIa; therefore, the conceptualisation of such vessels was more standardised in the late phases of the Geometric era. Neck-less trefoil oinochoai do not suggest any clear pattern at this stage and will be re-examined later.

The regression line for the same proportion of 6 Late Geometric pitchers in Figure 7 follows the equation y = 0.5507x – 0.2951 (where y = handle attachment height and x = net height). In other words, the handles of Attic LGII pitchers are attached at roughly 55% of a vessel’s net height, while the difference of 0.2951 cm is too small to be considered. The coefficient of determination of this regression line (R2=0.8429) suggests 84.29% statistical correlation, which is relatively satisfactory. This pattern could relate to a different technological tradition, where pitcher handles were attached roughly above the middle of a vessel’s height axis.

*The Kerameikos cemetery*

The analysis of metrical features (metrics) and proportions of medium sized pouring vessels from Kerameikos is conducted on 24 artefacts with complete profiles. The assemblage consists of 19 decorated oinochoai (one of which is neck-less), 1 oinochoe-lekythos and 4 pitchers. All vessels have been recovered in burials and their grave contexts are recorded in Charts 4 and 5.

In the original publication of oinochoai from Kerameikos, Karl Kübler (1954) recorded only height measurements. All other features in Charts 4 and 5 were measured in smaller scale through published photographs and then calculated in real scale based on the original real net height measurements. To ensure the accuracy of calculated measurements, 5 oinochoai from the Kerameikos assemblage were chosen for macroscopic analysis and an accuracy test was carried out similarly to that discussed in the earlier article on large containers (Chart 6). During this accuracy test, it was initially verified that the height measurements recorded by Kübler (1954) were correct. Then, real base diameters were obtained after macroscopic examination, which were compared to the ones calculated though published photographs for the same artefacts. According to Chart 6, the difference between real and calculated base diameters for the Kerameikos oinochoai ranges between -0.4 cm and +0.5 cm. Furthermore, differences between real and calculated proportions of base diameter to net height range between -1.7% and +2.6%. This test shows that differences between real and calculated metrical features exist; however, they are too small to affect the analysis results.

According to the correlation of base diameter and net height in Figure 8, the total assemblage of pouring vessels forms three distinct clusters. Cluster 1 comprises of 12 vessels from different chronological periods, ranging from EGII to LGIIb. All vessels have net heights below 25 cm. This cluster includes every Late Geometric oinochoe from the entire assemblage:

Cluster 2 consists of 6 oinochoai, all dating in periods before LGIa:

In this cluster, vessel heights range between 25 cm and 30 cm, while base diameters appear more diverse compared to those of cluster 1. Cluster 3 consists of all pitchers:

This cluster is the most distinct of all. Vessels are taller than 30 cm and their base diameters do not diversify. All pitchers date in LGII. Similarly to the pitchers from Kynosarges and the collections of the British School at Athens, pitchers from Kerameikos stand out with regard to their chronological and typological properties.

Finally, Figure 8 shows that two vessels are loners: firstly, oinochoe-lekythos 1141 (MGII) is the smallest of all artefacts. This vessel was originally added in the assemblage to test whether oinochoai-lekythoi cluster with other typological classes. Figure 8 suggests that such shapes stand out due to their broad bases and short heights. Secondly, oinochoe 2149 is the largest vessel in the entire assemblage and stands out even in relation to pitchers. Based on its height (40.7cm), 2149 could also classify as a giant trefoil oinochoe; however, the lack of distinct guidelines for the characterisation of such vessels and the tendency of scholars to see giant oinochoai as a LGI phenomenon (e.g. Brann 1961a; Galanakis 2013) prevent such characterisation for 2149. Instead, this vessel is treated as a standard trefoil oinochoe that happens to be larger compared to average pouring vessels from Kerameikos, and was also treated as such by Kübler (1954, 235). Similarly to the cluster with the largest neck-handled amphorae from Kerameikos and the cluster with the largest trefoil oinochoai from the Agora, 2149 has been produced in the period between EGII and MGI.

*Figure 8: Scatter-graph of correlation between net height and base diameter. Athenian oinochoai and pitchers from the Kerameikos cemetery with complete profiles.*

According to the above, all oinochoai produced after LGIa and all LGII pitchers show lesser degree of artefact variability compared to oinochoai produced before LGIa. Late Geometric vessels appear standardised and are likely to suggest the presence of distinct technological traditions in their production, similar to those discussed for the Agora assemblage.

The analysis of mean proportions across two broader chronological groups based on the data recorded in Chart 3 shows fluctuations in standard deviations. More specifically, the proportions of neck length to net height and base diameter to net height of oinochoai produced after LGIa are smaller compared to those before LGIa. By contrast, mean proportions of handle attachment height to net height show the opposite pattern:

Even though the total oinochoai assemblage from Kerameikos consists of 20 vessels (15 before and 5 after LGIa) the proportion of handle attachment height to net height could only be calculated for 17 vessels. This is due to the nature of some published photographs, which were taken from angles that prevent full visibility of the handles. Furthermore, mean proportions for pitchers show low standard deviations with exception of the proportion of handle attachment height to net height:

According to the above comparisons, it is likely that the neck lengths of Late Geometric oinochoai and pitchers were formed by potters based on specific conceptualisations, perhaps regulated by technological traditions. These resulted to lesser artefact variability, which is expressed in lower standard deviations in the above means. By contrast, the proportion of handle attachment height to net height follows the opposite pattern: oinochoai produced before LGIa exhibit smaller standard deviation and could be more standardised as opposed to those produced after LGIa. Both assumptions require further investigation.

Figure 9 plots the correlation between neck length and net height for 19 Geometric oinochoai with necks and 4 pitchers. By contrast to the material from the Agora, Kynosarges and the British School’s collections, pouring vessels from Kerameikos show different patterns of proportional increase between the two metrical features. Furthermore regression lines either follow unclear patterns or show limited statistical correlation.

*Figure 9: Scatter-graph of correlation between net height and neck length with regression lines. Athenian oinochoai and pitchers from the Kerameikos cemetery with complete profiles.*

More specifically, the regression line for trefoil oinochoai between EGII and MGII-LGIa follows the equation y = 0.4657x – 0.5297 (where y = neck length and x = net height). For this chronological group necks are roughly equal to 46.57% of a vessel’s net height, while the difference of 0.5297 cm is too small to be considered. For this specific assemblage the coefficient of determination of the regression line (R2=0.7778) shows weak statistical correlation at 77.78%.

According to the same scatter-graph, all trefoil oinochoai produced after LGIa appear closely clustered and no clear regression pattern is visible. Their regression line follows the equation y = 0.9888x – 13.705 (where y = neck length and x = net height), which is hard to explain. It could mean that neck lengths of Late Geometric oinochoai are roughly equal (99%) to their net height, reduced by an average of 13.7 cm. The coefficient of determination of this regression line (R2=0.507) shows weak statistical correlation (50.7%). This unclear pattern is most likely due to the sample’s nature and size.

A similarly unclear pattern is noted with regard to Late Geometric pitchers: their regression line follows the equation y = 0.7758x -15.013. This pattern could mean that neck lengths of pitchers are roughly equal to 77.58% of their net height, reduced by an average of 15 cm. The coefficient of determination of this regression line (R2=0.9937) shows perfect statistical correlation at 99.37%, which makes things more complicated to understand.

*Figure 10: Scatter-graph of correlation between net height and handle attachment height with regression lines. Athenian oinochoai and pitchers from the Kerameikos cemetery with complete profiles.*

The situation with the proportions of handle attachment height to net height for the Kerameikos assemblage is again problematic. According to the scatter-graph in Figure 10, all oinochoai produced after LGIa and all pitchers show regression lines that are difficult to interpret.

The regression line for Late Geometric oinochoai is y = -0.7884x + 29.244 and the coefficient of determination (R2=0.363) shows weak statistical correlation (36.3%). This equation does not change even if neck-less trefoil oinochoe 874 is left out of the data set. The same can be said with regard to pitchers, where the regression line follows the equation y = 0.0641x + 18.823, showing almost no statistical correlation (3.37% based on R2=0.0337). The only regression line that makes some sense is that for oinochoai produced before LGIa. According to the graph, their regression line follows the equation y = 0.4716x + 0.74. This group of oinochoai from Kerameikos has their handles attached at roughly 47.16% of a vessel’s net height, while the difference of 0.74 cm is too small to be considered. Furthermore, the coefficient of determination of this regression line (R2=0.8711) shows relatively satisfactory statistical correlation at 87.11%.

According to the above scatter graphs, the material from Kerameikos is problematic and no clear patterns are visible. The main problem is the small sample size and the dense clustering of all Late Geometric pouring vessels (both pitchers and oinochoai). This clustering is unlikely to verify patterns similar to the ones noted for the same proportions discussed for the Agora-Kynosarges-British School assemblage. A larger statistical sample of intact vessels is necessary to supplement the above study.

*The British Museum collections*

The analysis of metrical features (metrics) and proportions of medium sized pouring vessels from the collections of the British Museum is conducted on 23 artefacts with complete profiles. The assemblage consists of 10 decorated trefoil oinochoai (2 of which are giant), 1 oinochoe-lekythos, and 12 pitchers (recorded in Charts 7 and 8). All vessels derive from unknown contexts and are characterised as broadly Attic by Coldstream (2010), apart from those of suspected Athenian origin. Pitcher GR1877,1207.10 is probably from Phaleron and was added to test any similarities between Athens and other Attic fineware production centres.

In the British Museum assemblage, pitchers were the only vessels that could accommodate an accuracy test. For this test, published net height measurements were used to estimate rim diameters through photographs, which were then compared to real rim diameters included in the original publication by Coldstream (2010). According to the test (Chart 9), the difference between real and calculated rim diameters for these vessels ranges between -0.8 cm and +0.5 cm. The difference between real and calculated proportions of rim diameter to net height ranges between -1.9% and +1.6%. In Chart 9 pitcher GR1977,1211.4 could not allow clear calculation of rim diameter due to a large chip missing along its rim; therefore, this vessel was not included in the accuracy test. According to Chart 9, the differences between real and estimated metrical features are too small to affect the analysis results and bias is expected to be limited.

*Figure 11: Scatter-graph of correlation between net height and base diameter. Decorated Attic oinochoai and pitchers from the British Museum with complete profiles.*

Figure 11 plots the correlation of base diameter and net height for the above assemblage. According to that graph, both types of vessels are mixed and scattered in a different way compared to the assemblages discussed in the sections for the Agora and Kerameikos assemblages. Despite the high degree of scattering, pouring vessels form two clusters. Cluster 1 consists of 14 pots: 8 oinochoai, 1 oinochoe-lekythos and 5 pitchers (3 of which with short necks). The chronological range of this cluster spans over 200 years, as it dates between LPG and LGIIb:

The general characteristic of this cluster is that all vessels are shorter than 35 cm and their base diameters show greater variability compared to the second cluster. Cluster 2 consists of 9 vessels: 2 giant trefoil oinochoai and 7 pitchers:

This cluster is more homogeneous compared to cluster 1, as all vessels date between LGIb and LGIIb. Pottery in this cluster is taller than 35 cm, which is normal for pitchers and giant oinochoai.

Figure 11 suggests a trend that was not observed earlier in the analysis of the Agora-Kynosarges-British School and Kerameikos assemblages: even though oinochoai and pitchers from previous sites diversified clearly at heights of 33 cm and 30 cm respectively, pitchers from the British Museum appear in smaller heights and mix together with oinochoai. This makes it difficult to see clear artefact variability patterns.

The analysis of mean proportions across two broader chronological groups based on the information presented in Chart 8 shows that standard deviations for all of oinochoai produced after LGIa are smaller compared to those produced before LGIa:

Even though all oinochoai from the British Museum number 11 vessels in total (6 before and 5 after LGIa) the proportion of handle attachment height to net height could only be calculated for 9 vessels. This is due to the nature two published photographs, which were taken from angles that prevent full visibility of handles. By contrast to previous comparisons of mean proportions for pitcher, this assemblage shows high standard deviations that are due to high variability within the statistical sample:

According to the above comparisons, it is likely that neck lengths and handle attachment heights of oinochoai produced after LGIa were formed on specific conceptions, which were not necessarily followed before LGIa. The lower standard deviation in the above means suggests lesser artefact variability. Both assumptions are investigated below.

Figure 12 presents the correlation on neck length to net height for all oinochoai and pitchers from the British Museum. By contrast to the unclear patterns from Kerameikos, this assemblage follows similar proportional patterns to those observed for the Agora, Kynosarges and British School vessels. Furthermore, all regression lines appear parallel and two of them merge almost completely. Oinochoe GR1877,1207.10 is a loner which stands out. This specific vessel probably comes from Phaleron; therefore, it may be the product of a different chaîne opératoire which did not follow the technological patterns seen in the production of Athenian or other Attic pouring vessels.

*Figure 12: Scatter-graph of correlation between net height and neck length with regression lines. Attic oinochoai and pitchers from the British Museum with complete profiles.*

According Figure 12, the regression line for trefoil oinochoai before LGIa follows the equation y = 0.4153x – 0.5853 (where y = neck length and x = net height). This means that neck lengths of early trefoil oinochoai are roughly equal to 41.53% of a vessel’s net height, while the difference of 0.5853 cm is too small to be considered. For this specific assemblage the coefficient of determination (R2=0.5199) shows weak statistical correlation (51.99%). The same regression line for oinochoai produced after LGIa shows a clearer pattern. The regression line follows the equation y = 0.396x – 1.1538, which means that the necks of Late Geometric trefoil oinochoai are roughly 39.6% of a vessel’s net height reduced by 1.15 cm. The coefficient of determination of this regression line (R2=0.99) shows perfect statistical correlation (99%).

All pitchers from the same assemblage follow a similar regression line with Late Geometric oinochoai and both lines merge almost completely. The regression line for pitchers follows the equation y = 0.3857x – 0.7614 (where y = neck length and x = net height). The coefficient of determination of this regression (R2=0.7828) shows 78.28% statistical correlation, which is not entirely satisfactory. As all pitchers date in the Late Geometric era, the merging of their regression line with that for LG oinochoai could imply the presence of shared conceptualisations in the production of both wares: their necks were formed roughly at 40% (or 2/5) of a vessel’s net height; however, this suggestion must be treated with caution due to the different statistical correlations of both regression lines.

*Figure 13: Scatter-graph of correlation between net height and handle attachment height with regression lines. Attic oinochoai and pitchers from the British Museum with complete profiles.*

Similar patterns are noted in Figure 13 with regard to the proportion of handle attachment height to net height. In this scatter-graph, all oinochoai produced before LGIa follow the regression line y = 0.4806x + 1.8076, while the coefficient of determination (R2=0.6386) shows weak statistical correlation (63.68%). However, the same regression line for oinochoai after LGIa follows the equation y = 0.6228x-1.3916 (where y = handle attachment height and x = net height). The coefficient of determination (R2=0.9619) shows strong statistical correlation (96.19%). According to this pattern, the handles of oinochoai produced after LGIa were attached at roughly 62.28% of a vessel’s net height reduced by roughly 1.4 cm.

The regression line for Late Geometric pitchers in Figure 13 almost merges with that of oinochoai from the same period. Their regression line follows the equation y = 0.6508x – 2.6709, which means that vessel handles were attached at roughly 65% of a vessel’s net height reduced by an average of 2.67cm. The coefficient of determination of this regression line (R2=0.9448) shows strong statistical correlation (94.48%). It is likely that the handles of Late Geometric pitchers and oinochoai were attached at similar heights, a little below 2/3 (or 66.67%) of a vessel’s net height.

According to the above, the assemblage from the British Museum verifies the smaller degree of artefact variability of Late Geometric pouring vessels, as opposed to the Early and Middle Geometric ones. Furthermore, there are strong indications that the conceptualisation of Late Geometric oinochoai and pitchers followed similar proportional patterns. This is more evident with regard to the proportion of handle attachment height to net height, where the statistical correlation of the regression lines of both vessel types is high. In general, the British Museum assemblage exhibits more similarities with the Agora-Kynosarges-British School assemblage, as opposed to that from Kerameikos.

### Analysis of fabrics

Fabric analysis is conducted on 25 Athenian oinochoai from the Agora, summarised in Chart 10. According to hand specimen examination, the fabrics encountered in this assemblage are the same as the ones described for large sized containers. All vessels have been produced from the same fabric, which comes in two similar variants. Variant 1 is finer and harder, while variant 2 is softer, relatively ‘coarser’ and more calcareous (same as for large ceramic containers). By contrast to large ceramic containers, almost all Athenian oinochoai were produced from variant 1 and their Munsell (1975) colours mainly belong to the upper 5YR series. The most prevailing fracture colour is 5YR 5/4. There is only one sample, P18618, which was produced from variant 2 and dates in EGII.

*Figure 14: Fabric division for 25 oinochoai from the Athenian Agora.*

According to Figure 14, variant 1 comprises 96% of the examined material. This fabric was used for the production of various oinochoai regardless of the period of their production and context of recovery. The use of a single fabric is again indicative of a strong technological tradition in clay selection, manipulation and tempering practices.

### Analysis of decorative technology

The analysis of decorative technology is conducted on a total of 56 pouring vessels (both complete pots and sherds) from three sites. The assemblage from the Athenian Agora consists of 47 decorated oinochoai (Chart 11). The assemblage from the Kynosarges burials consists of 1 oinochoe and 1 pitcher, and the assemblage from the British School’s collections consists of 2 oinochoai and 5 pitchers. The latter two assemblages are examined together due to their small size (Chart 12).

*The Athenian Agora*

Decorative colours of oinochoai from the Athenian Agora follow the same pattern observed for large sized containers from the same context . According to Figure 14, 94% of the samples dating before LGIa are painted with colours of Group 1 (black or brownish black). The same colours prevail in the period after LGIa (93%); however, next to them there appear colours of Group 2 (brownish red or red) at 4% and Group 3 (orange or reddish yellow) at 3%. According to the comparison, colours for the decoration of oinochoai were highly standardised across time with only few exceptions. Colour Group 3 should be treated as a Late Geometric phenomenon with limited presence.

*Figure 14: Comparison of decorative element colours of Athenian decorated oinochoai from the Agora.*

Additionally, the period between EGI and MGII-LGIa is characterised by the dominance of coated vessels, similarly to the case of closed ceramic containers. According to Figure 15, vessels produced with black or brown/black coated surfaces (colour Group 1) consist 89% of the assemblage. Red or brownish red coated oinochoai (Group 2) consist 5% of the assemblage and only 6% of all vessels are uncoated. By contrast, after LGIa the majority of the samples (62%) are uncoated and the vessels coated in colours of Group 1 drop down to 38%. During the same period red and brownish red coatings (colour Group 2) disappear.

*Figure 15: Comparison of coating colours of Athenian decorated oinochoai from the Agora.*

*Figure 16: Slip or coating quality of Athenian decorated oinochoai from the Agora.*

Despite the decline of coating practices after LGIa, Figure 16 shows that the quality of external treatments of the Late Geometric period is more diverse: it includes thin lustrous coatings and thick metallic sheens, which do not exist in the earlier assemblage.

The above analysis suggests that Late Geometric oinochoai show greater artefact variability compared to those produced between EGII and MGII-LGIa with regard to their decorative characteristics. Late Geometric vessels are by majority uncoated and this could be due to the spread of the figurative style after c.760 BC, which required larger ‘blank’ surfaces for the painters to work on. The presence of more than one decorative colours and the existence of different qualities of coatings during the same period shows advances in the preparation of paints resulting to multiple chemical compositions, also related to advances in firing control. Similarly to elaborately decorated amphorae discussed in Chapter 4, the practice of coating the external surfaces of oinochoai declined significantly in Late Geometric times but it was never abandoned completely.

### The Kynosarges burials and the collections of the British School at Athens.

The assemblage of oinochoai from those two sites is small to produce certain conclusions; however, according to the information presented in Chart 12, the assemblage probably complies with the patterns observed for oinochoai from the Agora with regard to their coatings and decorative colours. The decoration of Late Geometric pitchers appears to be homogeneous. All pots presented in Chart 12 are painted with motifs in colours of Group 1 and all vessels are covered with a thin matte wash in the colour of the original clay. The decoration and external treatment of pitchers follows the same characteristics described earlier for Late Geometric oinochoai.

### Summary and discussion of analysis of pouring vessels

The analysis of artefact variability of trefoil oinochoai and pitchers suggests that both pouring vessels bear technological similarities and must be treated as products of the same chaîne opératoire. Even though standard trefoil oinochoai were produced all along the 9th and 8th centuries BC, neck-less, broad and giant trefoil oinochoai, and pitchers appeared for the first time during the Late Geometric period. Neck-less trefoil oinochoai were the products of a simplified chaîne opératoire with fewer steps compared to that of standard trefoil oinochoai. Their production was probably meant to cover increasing consumption demands for pouring vessels during Late Geometric times. Still, neck-less vessels bear the same properties with every other pouring vessels of that time. Pitchers and giant trefoil oinochoai were produced in larger sizes compared to standard, neck-less and broad trefoil oinochoai. Furthermore, pitchers exhibited greater standardisation with regard to their metrical features and decorative characteristics compared to all other typologies. There are some indications that the potters who produced pitchers had similar conceptions with those who produced all other pouring vessels, particularly in the shaping of their necks and the attachment of their handles.

The analysis of metrical features and proportions in this chapter shows that there is a chronological boundary in the production of Geometric trefoil oinochoai, set at the beginning of LGIa. Early and Middle Geometric vessels exhibit greater artefact variability compared to Late Geometric, which cluster closely with regard to their net height and base diameter measurements. By contrast to amphorae, technological traditions in the production of oinochoai were not the same across time. The strongest technological traditions in their chaîne opératoire were most likely established in the Late Geometric period.

The regression lines for the proportion of neck length to net height for oinochoai produced before LGIa show that necks range between 35% and 47% of a vessel’s net height; however, all regression lines show weak statistical correlation:

*Figure 17: Scatter-graph of correlation between net height and neck length. Athenian/Attic pouring vessels produced before LGIa from the Agora, Kynosarges, Kerameikos, British School at Athens and British Museum collections with complete profiles.*

Comparing assemblages that were measured with different techniques (e.g. artefact handling as opposed to published photographs) could be relatively problematic. Still, the scatter-graph for the proportion of neck length to net height for a total of 35 vessels with necks from all sites in Figure 17 verifies the above degree of uncertainly. According to the graph, Early and Middle Geometric oinochoai with necks appear highly scattered and their regression line follows the equation y = 0.4002x + 0.1697. In other words, oinochoai necks were produced at an average of 40% in relation to a vessel’s net height. The coefficient or determination of this regression line (R2=0.6301) shows weak statistical correlation (63.01%).

The comparison of the same proportions for oinochoai produced after LGIa is confusing due to nature of the Kerameikos assemblage. The regression for Kerameikos makes no particular sense due to the nature of the sample:

Still, the regressions from the Agora-Kynosarges-British School and British Museum assemblages suggest that the necks of such oinochoai were produced at a proportion between 36% and 40% of a vessel’s net height. This range is smaller compared to that for Early and Middle Geometric vessels. The regression for the assemblage from the British Museum shows almost perfect statistical correlation, which could indicate a distinct technological tradition.

The same thing is noted with regard to the same proportion for pitchers, which all come from the Late Geometric era. There, regression lines show that neck lengths range between 38% and 42% in relation to a vessel’s net height, even though the pattern for the Kerameikos assemblage does not match:

Although the regression from Kerameikos shows perfect statistical correlation, it makes no sense due to the nature of this assemblage. Furthermore, the coefficients of determination for the other two regression lines show ambivalent statistical correlation.

Despite this unclear situation, the correlation of neck length to net height for a total of 17 oinochoai with necks and 22 pitchers produced after LGIa from all five sites in Figure 18 suggests at least one technological tradition with high certainty. According to the scatter-graph, the regression line for oinochoai follows the equation y = 0.3822x – 0.6189. In other words, necks of LG oinochoai were formed roughly 38.22% of a vessel’s net height, while the difference of 0.6189 cm too small to be considered. The coefficient of determination of this regression line (R2=0.9555) shows strong statistical correlation (95.55%).

*Figure 18: Scatter-graph of correlation between net height and neck length. Athenian/Attic pouring vessels produced after LGIa from the Agora, Kynosarges, Kerameikos, British School at Athens and British Museum collections with complete profiles.*

The same proportion for pitchers in Figure 18 follows the equation y = 0.4051x – 1.0496 and partly merges with the regression line of oinochoai. In other words, the necks of these vessels were formed at roughly 40.51% of a vessel’s net height reduced by 1 cm. The coefficient of determination of the regression line (R2=0.7739) does not show satisfactory statistical correlation (77.39%). It is interesting that the only pitcher suspected to be from Phaleron stands out. This vessel is most likely the product of a different workshop compared to all other vessels.

The identification of distinct patterns related to the proportion of handle attachment height to net height of pouring vessels is clearer compared to their proportion of neck length to net height. Regression lines for oinochoai produced before LGIa show that handles were attached at heights between 47% and 60% of a vessel’s net height; therefore, they suggest high degree of artefact variability:

According to the coefficients of determination, the assemblages from the Agora-Kynosarges-British School and Kerameikos show relatively satisfactory statistical correlations; however, the regression line from the British Museum assemblage must be treated with caution.

*Figure 19: Scatter-graph of correlation between net height and handle attachment height. Athenian/Attic pouring vessels produced before LGIa from the Agora, Kynosarges, Kerameikos, British School at Athens and British Museum collections with complete profiles.*

The scatter-graph for the proportion of handle attachment height to net height for a total of 31 EG and MG oinochoai with recorded handle attachments in Figure 19 verifies the above degree of uncertainly. According to the graph, Early and Middle Geometric oinochoai appear to be relatively scattered and their regression line follows the equation y = 0.5526x - 0.4307. In other words, the handles of oinochoai were attached at roughly 55.26% of a vessel’s net height, while the difference of 0.4307 cm is too small to be considered. The coefficient or determination of this regression line (R2=0.7864) shows weak statistical correlation.

By contrast, the situation with oinochoai produced after LGIa is different. Regression lines for standard trefoil oinochoai from the Agora-Kynosarges-British School and British Museum assemblages show high statistical correlation (above 96%), and therefore, low degree of artefact variability. The handles of such oinochoai were attached between 62% and 66% of a vessel’s net height:

The Kerameikos assemblage and all neck-less vessels from the Agora-Kynosarges-British School assemblage stand out: their regression lines make little sense and their statistical correlation is weak.

The same phenomenon is noted with regard to Late Geometric pitchers. Regression lines from all assemblages except Kerameikos suggest variations in handle attachment heights that range between 55% and 65%. Statistical correlation varies: it is relatively satisfactory for the Agora-Kynosarges-British School assemblage and strong for the British Museum assemblage:

Figure 19 plots the proportion of handle attachment height to net height for a total of 18 oinochoai and 21 pitchers with recorded handle attachments from all sites. The scatter-graph verifies that vessels produced after LGIa exhibit clear regression patterns, parallel alignment and high degrees of statistical correlation. According to the graph, Late Geometric oinochoai form a regression line that follows the equation y = 0.6042x - 0.3899. In other words, the handles of oinochoai produced after LGIa were attached at roughly 60% of a vessel’s net height, while the difference of 0.3899 cm is too small to be considered. The coefficient or determination of this regression line (R2=0.9491) shows strong statistical correlation (94.91%).

The same regression line for Late Geometric pitchers follows the equation y = 0.6156x - 2.0422 and runs parallel to that of oinochoai. This means that the handles of pitchers were attached at roughly 61.56% of a vessel’s net height, reduced by an average of 2 cm. The coefficient of determination of this regression line (R2=0.8998) shows relatively strong statistical correlation (90%). Once again, the pitcher suspected to come from Phaleron is a loner.

*Figure 19: Scatter-graph of correlation between net height and handle attachment height. Athenian/Attic pouring vessels produced after LGIa from the Agora, Kynosarges, Kerameikos, British School at Athens and British Museum collections with complete profiles.*

According to the above comparisons of two basic proportions, it is clear that Late Geometric pouring vessels were more standardised compared to their Early and Middle Geometric counterparts. The lesser degree of artefact variability of pottery produced after LGIa is likely to suggest a small number of workshops or artisans involved in the production of pouring vessels. By contrast, the higher degree of artefact variability during the earlier Geometric phases suggests more workshops and greater diversity. Although some distinct patterns for conceptualising different constituent vessel parts might have existed among potters before LGIa, these never became a tradition in the same sense that this was noted for amphorae. Artefact variability continued after LGIa; however the lesser degree of scattered variables and the relatively higher degree of statistical correlation of regression patterns suggest the establishment of specific technological traditions in vessel conceptualisation during Late Geometric times. The neck lengths of standard trefoil oinochoai were formed to be roughly 37.5% of a vessel’s net height, or in other words roughly shorter that 2/5 (40%). The same conception might have existed for pitchers, although statistical correlation is questionable. Secondly, the handles of all Late Geometric oinochoai (regardless typological class) were consciously attached at roughly (60%) of a vessel’s height axis, or in other words at roughly 3/5. The same proportion (roughly 61%) was also followed in the conceptualisation of pitchers. The statistical correlations of the regressions related to handle attachment heights for oinochoai and pitchers are likely to suggest a strong technological tradition, followed by almost every Late Geometric workshop.

The Late Geometric material without context coming from the British Museum and the collections of the British School at Athens shows great similarities with that from the Agora, Kerameikos and Kynosarges. It is more than likely that all samples from the British Museum and the British School are not only Athenian, but also produced at the same site as all other pouring vessels. According to Papadopoulos (2003) this production site matches the Agora. Furthermore, if a small group of Athenian potters or workshops were responsible for the entire production of Late Geometric pouring vessels, it is likely that this production monopolised a large portion of the broader Attic market. The pitchers coming from the British Museum are a good example in support of this point: all of them exhibit great similarities with the Kerameikos and Kynosarges clusters, with exception of GR1877,1207.10, which is probably the product of a Phaleron workshop.

The strongest technological tradition noted in the production of pouring vessels (and more specifically oinochoai) relates to the use of a single fabric all across two centuries. Hand specimen examination reveals that the majority of oinochoai from the Athenian Agora were produced out of the same clay as amphorae (variant 1).

The dominant colours used for decorative elements and coatings of pouring vessels all across the Geometric era were black and brown black (Colour Group 1). The period between EGI and MGII-LGIa was characterised by a general use of thick lustrous or matte coatings on the external surfaces of oinochoai. This practice began to decline after LGIa and was gradually replaced by a preference in thin matte slips or plain washes in the colour of the original clay. This easier way of finishing vessels was most likely related to the spread of the figurative style in pottery decoration. Still, thick-coating was never abandoned completely and continued to be practised all along the Late Geometric period by exhibiting patterns of greater variability in quality by contrast to earlier times. Such patterns are likely to suggest that there were larger numbers of painters involved in the decoration of Late Geometric pouring vessels compared to the potters who manufactured them. Painters practised the same coating techniques as those from earlier times and also enriched them with innovative ideas (e.g. the metallic effect of highly lustrous sheens). By contrast, the decoration of pitchers was standardised: all vessels were painted with brown or black motifs and none carried coated surfaces.

According to the above conclusions, it is highly likely that the production of pouring vessels faced significant changes sometime at the beginning of LGIa. The shapes of oinochoai became more standardised, probably as a result of fewer workshops or artisans involved in their production. At the same time, neck-less, wide and giant oinochoai, and pitchers appeared alongside as distinct shapes, yet produced with the same conceptualisations as standard trefoil oinochoai. Despite the lower degree of artefact variability with regard to their metrical features and proportions, Late Geometric oinochoai exhibited higher variability with regard to their external treatments and decoration. The most possible explanation is that despite the reduction in the numbers of potters or workshops involved in the shaping of such pots, their decoration passed to the hands of a larger number of artisans. Some of these painters were highly experienced in elaborate coating techniques. In that sense, oinochoai are the best vessel class to demonstrate that the work of painters and potters was separate, at least after LGIa.