Cite this as: Baddiley. W. 2018 Calculating Liquid Capacity to Understand what could have been Consumed from 'Drinking' Vessels, Internet Archaeology 50. https://doi.org/10.11141/ia.50.4
This article discusses the calculation of drinking vessel capacities and what these can tell us about consumption practices. The principal evidence considered consists of the capacities of the pottery 'drinking' vessels from the legionary fortress at Usk, south-east Wales (Greene 1979; 1993), which are compared with those of glassware found in Colchester (Cool and Price 1994), and silverware from the House of the Menander in Pompeii (Painter 2001). Differences or changes in vessel capacities within these distinct groups, and comparison of different vessel materials, can help indicate different consumption practices.
Other studies of Roman vessels that have calculated vessel capacity to aid the interpretation of vessel use, including Dannell (2006) and Hudson (2010), have informed the calculation of capacity used in this article. However, the research reported here differs by examining the capacities of a specific range of vessels and vessel materials that could have been used for consumption, and especially liquid consumption. Traditionally, consumption has been a neglected topic within archaeological research, which has focused on production (Allison 2013, 35-36). Knowing that a certain product was produced and transported to a site can be useful, but finding out how, where and why it was used, or consumed, can potentially offer a greater insight into the ancient world.
Determining the type of liquid held in the vessels can be approached by examining volume calculations. Such capacities that are calculated can only be used as a guide, however, since we do not know the levels to which the vessels were filled, and while there tends to be an inverse correlation in the modern world between alcoholic strength and vessel size, it cannot be assumed that only alcohol was being consumed from these Roman-period vessels. Hot drinks or water could be consumed from them, or they could have been used for dips or for foodstuffs such as garum (see e.g. Dannell 2006, 160). There was also the tendency to mix wine and water, in varying ratios (Dunbabin 2003, 20, 22), which means the volume has an even more imprecise relationship to alcoholic strength.
Two capacities are calculated in this article: the optimum fill (Opt) and the maximum fill (Max). My approach has been influenced, in particular, by Dannell's study of samian cups (2006) and by Hudson's study of the archaeology of the Roman convivium (2010). Dannell calculated the capacities of samian ware by measuring to a 'maximum fill-height' using the internal grooves found on cups (2006, 157, 167). This approach does not factor in the entirety of the vessel; rather it measures to a more realistic 'optimum' fill level, and forms the basis of my calculation of the Opt. Hudson used the method of 'Summed cylinders' to calculate capacity (2010, 677). In effect, this creates a series of small cylinders within the vessel which are then combined for the final volume. This 'Summed cylinders' approach factors in the entirety of the vessel, which is reflected here in my calculation of the Max.
The formula for calculating the volume of a cylinder V=Πr²h is used here as the basis for the calculation of capacity. Should the vessels have a different width at the top and bottom, a variation of the formula is used:
, where D1 and D2 are the values at the top and bottom of the cylinder. Vessels that have been illustrated for publication with a complete line drawing are suitable for analysis by the methods described here. Such drawings show the inner vessel surface needed to calculate vessel capacity. Consequently some vessel forms are represented by only one or two vessels (see section 2.4) while other forms are represented by many vessels (see section 2.3). The images of the Menander silverware do not show the inner linings of the vessels, which does not allow for as accurate a calculation, as can be seen in the large volume differences between Opt and Max (Table 2). This does not mean that the capacity of the vessels cannot be calculated, just that the final volumes need to be treated as slightly less accurate than the capacities for the other vessels.
Drag. 27 no. 145 from Usk is used as an example to demonstrate how the Opt (Figure 1) is calculated here. The Max is calculated in the same way but the entirety of the vessel is factored in. The height of each section that the vessel is split into is kept small, ideally between 5-10mm, to preserve the accuracy of the calculation and allow more of the vessel to be included in the calculation. This process gives results that will be approximately correct, because the calculation produces an average radius for each section, in effect creating a series of cylindrical 'steps' (Figure 2). The calculated volumes presented and discussed here have been rounded to the nearest 10ml, and any vessels with a capacity under 100ml have been excluded.
I have classified vessels used in this study into four categories: cups, beakers, flagons and jugs. These vessel forms seem likely to have had the most pronounced use for liquid consumption. However, to compare data from different archaeological project reports, I needed a common classification system to compare vessels. I have used Webster's guide (1969) to coarse pottery as the template for naming and describing vessel forms.
Webster did not define cups, but I have used this label for handheld vessels that are wider than they are tall. The ratios of the Usk pottery cluster around 2:1 (width:height), but I have considered a vessel anywhere between ≥1:1 and ≤3:1 to be a cup. Beakers are also handheld vessels (Webster 1969, 9), but they differ from cups in that they are taller than they are wide. The beakers found at Usk cluster around the 1:1.3 ratio, so for the purposes of classification I have considered vessels with a ratio of 1:1-1:2 as beakers. The differences between jars and beakers cannot be seen in the ratios, since they tend to be of a similar size. Jars are typically identified as storage or cooking vessels, not consumption vessels. A jar would have a constriction at the neck to allow it to be sealed (Webster 1969, 10), and would fall in the 1:1-1:2 ratios. Where there is no pronounced constriction at the neck I have considered the vessel to be a beaker.
|Vessel form no.||Greene classification||New classification|
Webster (1969, 9) defined a flagon as a 'vessel for holding liquids, with a narrow neck, usually globular body … and one or more handles'. The handle, or multiple handles, on flagons could also imply that they were used for pouring. The size of the flagons at Usk tends to vary, so I have interpreted vessels with ratios over 1:2, globular bodies and at least one handle as flagons. Webster (1969, 10) interpreted spouted flagons or jars/beakers as jugs. These fall in the ratio of 1:1 and more.
Greene assigned four individual vessels in the Usk archaeological reports to dual, or alternative, classifications (i.e. flagon/jug, jar/beaker), which do not concur with my more specific classification criteria (Figure 3). I have therefore changed Greene's original classifications, according to my criteria and based on their respective ratios, so that each of these particular vessel types has only one identifying classification (Table 1). This allows all of the pottery vessels in this study to be assessed by the same classification criteria. The ratios for the vessel form types U7, U14a and LW12.3 concur well with my criteria but those for vessel form type U14b does not. Therefore vessel form types U7, U14a and LW12.3 are all types of vessel that can be analysed below, but form type U14b does not fit into my classification criteria and so is left out of the analysis.
The approach used for pottery vessel forms is not suitable for classifying the silverware from the House of the Menander in Pompeii. Using my pottery-based system to reclassify and rename these silverware vessels would only cause confusion. For example, applying the pottery-based ratios to silverware vessels M5-M8 would classify them as beakers, despite the fact that they have two handles and more in common visually (Figure 4) with the rest of the silverware than with other vessels identified as beakers. Therefore, I have used Painter's original vessel labels (2001, 18, table 3.1, 62-63) to classify these forms (Table 2). However, for the purposes of the analysis carried out here, I will treat M1-M10 as two-handled cups, M11-M13 as one-handled cups and M25-M26 as beakers based upon their appearance. This is to highlight that the silverware vessels, M1 to M10 in particular, differ from the pottery vessels because of the handles on the majority of the vessels (Figure 4) as well as the short stems on several (M5-M8). This difference in classification, however, does not affect the function of the vessels since these are well attested as drinking vessels and for wine consumption in particular (Painter 2001, 32). The silverware vessels will be assessed alongside the pottery and glass vessels, but will be kept separate.
|Opt (group)||Max (group)|
|Group||Painter classification||Cat. no.||Opt (ml)||Max (ml)||Range (ml)||Mean (ml)||DFM||Range (ml)||Mean (ml)||DFM|
Using the above calculation and classification criteria, the capacities of the vessels from Usk, Pompeii and Colchester are discussed individually to highlight any patterns in their capacities, before all of the vessel capacities are compared.
The pottery vessels from Usk come from a limited period of time within the first century CE. As well as the usual examples of imported pottery, like Lyon ware and samian ware, the collection includes a number of unique, locally made vessels: the Usk ware. Analysing this material presents an opportunity to see what the soldiers occupying the legionary fortress preferred, since the vessels were made for them in the fortress kilns. The Usk pottery data is then compared to the silverware found at the House of the Menander at Pompeii. These two sites have the benefit of being roughly contemporary with one another: Usk was in use between 49 and 74 CE, while the site at Pompeii was preserved by the eruption of Vesuvius in 79 CE. Four first-century CE glassware vessels from Colchester are also included to see how the capacities of glass vessels compare to those of the pottery and silver vessels. The datasets for the Lyon ware and Usk ware comprise vessel form types, while the Drag. 27, Colchester glassware and Menander silverware are composed of individual vessels. The reasons for this are outlined below.
For the purpose of this study, 'communal consumption' means consumption from one vessel shared between several people, while 'individual consumption' refers to each person having a separate vessel to drink from. Whether or not a vessel is thought to have been used for individual or communal consumption tends to be based on vessel size. Cool notes that beakers from King Harry Lane cemetery, St Albans, England, would possibly have been communal vessels because of their large size (2006, 164), and Wallace also mentions the use of physical characteristics, forms and sizes as a way of gauging the possible use of vessels, with 'small vessel diameters and volumes … likely to have been for individual use' (2014, 107). Luley (this volume) mentions that vessels with diameters larger than 17-18cm or smaller than 6-7cm are unlikely to have been used for drinking. For this article, I assume that vessels with a small capacity (<1L) and larger pouring vessels (>1L) were for individual consumption and that large direct consumption vessels (>1L) were for communal consumption. I have chosen boundary points based on capacity and not diameter for several reasons. As noted below, the use of vessels with a capacity of less than 1L for drinking is contested. The first reason is that for the Menander silver, for example, there is a noticeable gap of nearly 500ml between the largest vessel under 1 litre (M1 – 620ml) and the smallest vessel over 1 litre (M21 – 1100ml), so this boundary serves to distinguish clearly between 'small' vessels and 'large' vessels. The second reason for choosing these boundary points is that drinking vessels with a capacity of over 1 litre would seem to be more suited to several drinkers, rather than just one. This is reflected in the cut off of more than 1 litre for direct consumption vessels, which would indicate several drinkers and therefore communal consumption.
The silverware from the House of the Menander at Pompeii differs from the vessels found at Usk in several ways, not least the vessel material. The original drawings (Painter 2001, 46-47, figs 2 and 3) and see Figure 4 here, used to calculate the capacity of these silver vessels, are different to the line drawings of the pottery and glassware from Usk and Colchester in that they do not show the inner surfaces of the vessels, which does not allow for as accurate a calculation.
There are seventeen drinking vessels among the Menander silver, fourteen of which Painter grouped into pairs according to appearance and which can also be seen to have similar capacities (Table 2 and Figure 4). The DFM (Difference From Mean) columns in Table 2 are used to show how similar the vessels within Painter's groups are to each other. These columns refer to the percentage difference from the mean capacity for the group, calculated using:
The same calculation is used in Table 4 and Table 5 to show variation within the datasets. There are three vessels (M13, M20 and M21) that Painter does not put into a group. Using the capacity data that I have calculated, these three vessels are still not suitable for grouping together: Jugs M20 and M21 are the largest jugs in the dataset but I have left them ungrouped because there is a variation of 820ml in their maximum capacities, which is too large a divide to group them together; and Beaker M13 does not have a comparable vessel. The cups have a broad range of capacities from 210ml to 620ml (Opt) and 220ml to 720ml (Max). There is very little variation within each group in Opt, and slightly more in Max with an 80ml range in Group 6 and 90ml in Group 2, but at no point do vessels in a particular group differ more than 10% from the mean.
By the second half of the first century CE, glass vessels had ceased to be luxury items, and the glass industries had begun to make vessels for everyday and utilitarian use (Cool and Price 1994, 211-12). However, there is evidence for high-status use of glass vessels on an everyday basis at Alstone Cottage, Caerleon in south-east Wales (Price 1995, 80), and similarly at Segontium in north Wales (Casey and Davies 1993, 79-80), both of which date to the mid- to late first century CE. At Colchester, which was the legionary fortress of the XX Legion in the mid-first century CE, and a colonia from c. 49 CE, glass vessels seem to follow the trend of utilitarian use as they were found in greater numbers in the soldiers' barracks than in the centurions' quarters (Cool 2006, 177-78). The glassware from Colchester is found in contexts dating from c. 40 CE through to the fifth century CE (Cool and Price 1994, 211), but only the glassware from the first century CE is examined here. Cool and Price considered glass cups and beakers to be drinking vessels, while bowls have more of a culinary aspect (1994, 222). The classification criteria used for the vessels examined in this study (see above) are in line with Cool and Price's observations.
Capacities were calculated for four examples of glass drinking vessels from Colchester (Figure 5). According to Hilary Cool (pers. comm. 24 Feb. 17) these drawings are not very accurate representations of the actual vessels. Despite this, these images are still usable for my purposes here, because unless the original drawings were very inaccurate there would be little impact on the capacity calculations. The three glassware cups (nos 232, 290, 553) are larger in capacity than the only beaker (no. 426), but none of the vessels has a particularly large capacity (see Table 3). These are the only glass vessels from Colchester dating to the first century CE that have sufficiently complete line drawings in the archaeological report to enable them to be used to calculate capacities using the methods described here (Cool and Price 1994, 49, 67, 81, 91). There is ample evidence that the forms of the Colchester glassware varied over time, and even for the restricted time examined here there should be at least six, or as many as eleven, different forms represented for drinking vessels (Cool and Price 1994, 211-23), not just four vessels.
|Classification||Cat. no.||Opt (ml)||Max (ml)|
Drag. 27 vessels comprise the largest dataset of a single vessel form type and fabric recorded from Usk. I have therefore examined them separately here to demonstrate that even within one specific type of vessel there is a large degree of variation in capacity. These Drag. 27 cups demonstrate two distinct size groups: one group of smaller vessels and one of noticeably larger vessels. The smaller vessels (those under 100ml) are not discussed in this article. Instead, my analyses focus on the seventeen larger Drag. 27 cups (those with a capacity greater than 100ml – see Figure 6) because these are more likely to have been used for liquid consumption. Drag. 24/25 vessels were also present at Usk in groups of both small and large vessels, but are not included here because there was only one example of such a cup with a capacity in excess of 100ml. Also, the use of this form as a drinking cup is contested (see Dannell 2006, 158-62).
|Type||Cat. no.||Opt (ml)||Max (ml)||Opt DFM||Max DFM|
|Opt (ml)||Max (ml)|
|Mean capacity of whole group||238.2||281.18|
|Difference between smallest and largest vessel||170||180|
I have used the numbering system from the archaeological report ('Cat. no.' in Table 4) to identify the individual Drag. 27 vessels discussed in this article. Among the seventeen larger Drag. 27 vessels from Usk considerable variation between Opt and Max capacities can be seen (Table 4). In particular, the DFM values indicate that two vessels (nos 55 and 104) have large positive percentage differences from the mean and two other vessels (nos 135 and 143) have large negative differences. Size variation among these vessels is not unexpected, since they were not produced to the exacting standards of the modern production line. What is surprising, though, is the degree of variation.
|Type||Cat. no.||Opt (ml)||Max (ml)||Opt DFM||Max DFM|
|Opt (ml)||Max (ml)|
|Mean capacity of whole group||233||278|
|Difference between smallest and largest vessel||80||90|
To eliminate the effect of this variation on the analyses, I have removed these four larger vessels. With these outliers removed the DFM values have all changed and are more evenly spread (Table 5). The ranges of Opt and Max capacities are each reduced by 90ml. This pronounced change emphasises the effect that outliers can have on a dataset. After the removal of these outliers the remaining thirteen vessels have a much tighter range of capacities. But what is interesting is how little this affects the mean, with only a reduction from 281.18ml to 278ml in Max, and 238.2ml to 233ml in Opt.
The capacities of the Drag. 27 cups vary by a considerable amount: the largest vessels (55 and 104) have twice the capacity of the smallest vessels (135 and 143) in Opt, and by 180ml in Max. This could be an indication of different potters making vessels to different size preferences. It could also imply that the vessels were made with a particular capacity or size in mind, since the average vessel capacity is not affected to a great extent by the removal of the four outliers from Table 5. However, this does indicate that there is considerable variation in the capacity of the individual vessels.
The majority of the colour-coated cups reaching Usk in the first century CE came from Lyon, France (Greene 1979, 139). Lyon ware beakers were also reaching Usk, and these have had capacities calculated alongside the Lyon ware cups (Table 6). Greene (1979, 18) has divided his main vessel form types for these cups and beakers into sub-types according to rim form. This is particularly noticeable for the vessels within type 1.2 (i.e. L1.2a and 1.2b) and type 1.4 (i.e. L1.4a and L1.4b) (see Figure 7). I have not analysed the Lyon ware cups and beakers by each form type in the same fashion as for the samian Drag. 27 because there are not enough vessels of any specific type for such analyses to be meaningful. Instead I have analysed all the Lyon ware vessels that were drawn in enough detail that I could calculate volume, which has led to some vessel form types being represented by only one vessel (for example L6 and L8) and others having several vessels within the form type (for example L5.2a and L5.2b). Certain features of these Lyon ware cups and beakers can be discussed with reference to the capacities of their various specific form types (Table 6).
|Baddiley reference||Classification||Greene form type||Greene figure||Opt (ml)||Max (ml)|
|L8||Cup||8||12.2||170||220||Opt (ml)||Max (ml)|
Only two of the Lyon ware cup form types (L3.1 and L5.2a) are larger than 300ml in Max capacities, and on the whole the cups tend to cluster around 200ml in Opt capacity. They tend to have slightly lower capacities than the Drag. 27 vessels (see Table 4 and Table 5): the mean capacity for the Lyon ware cups is roughly 30ml less than the mean for the Drag. 27. This puts all the cups together regardless of 'Form type' so one might assume that they were made with a similar mean capacity in mind. There is no evidence that this should be the case, though, since the form that is classed as a cup does not appear to have a direct relationship with the volume of liquid that it can hold. For example, the three instances of type 1.4 (L1.4a, L1.4b and L1.4c) vary by 70ml in Opt and 100ml in Max capacity, and the variation is even more pronounced in beakers of type 20 (L20.1, L20.4 and L20.5), which vary in capacity by over 1.5 litres. Comparing these different vessels and their capacities could be problematic, as can be seen by the variation within each type and sub-type.
There are thirteen vessel forms in the Usk ware series that can be associated with drinking (Figure 8). As noted above, Usk ware was made in the fortress kilns and must have been intended for the soldiers at the fortress and so offers a unique insight into the kind of vessels required or considered to be necessary. The Usk ware vessels are only represented here by one vessel per form, unlike the Lyon ware vessels, which have a number of vessels per form, and the Drag. 27, which is one form but with several vessels represented.
|Baddiley reference||Classification||Form type||Opt (ml)||Max (ml)|
The importance of analysing the capacities of these vessel form types and not just their size or functional classification can be seen quite clearly in the case of the vessel types I have classified as flagons (Table 7). These vary in Max capacity between 410ml and 5500ml, so it cannot be assumed that all such classified flagons were used in the same way(s) for the same function(s). The five smallest vessel form types (U3, U14, U23, U24 and U25) could have contained volumes of liquid that suggest their use by a single drinker. The seven largest vessel form types (U1, U2, U4, U5, U6, U7, and U8) present a somewhat different picture. From their flaring rims and the spout on jug U6, flagons U5 and U7 and jug U6 are each clearly intended to pour from. Flagon U8 is squat and two-handled with a large rim, and so could have been used to drink from directly (Greene 1993, 17). Flagons U1, U2 and U4 are most likely to have been used to pour from, but the position of the handles and the rim indicate that they could also have been consumed from directly. The large capacities of flagons U1 and U2 indicate that direct consumption would have been difficult, so it is more likely that they were used to pour from.
The forms in which this type of pottery occurs would have reflected the needs of the soldiers since the fortress kilns would have made what was needed, or at least what was in demand. Drinking vessels make up only 17% of the total Usk ware assemblage (Greene 1993, 10) and, within this, the percentages of the different vessel types are particularly interesting. Flagons comprise 11% of the drinking vessel assemblage, with over a quarter of the flagon assemblage being two-handled type U8 vessels, while the cups comprise 2.11% of the drinking vessel assemblage (Greene 1993, 8-10). Thus, there are more of the type U8 flagons than all of the Usk ware cups put together. While this could indicate that the soldiers at Usk preferred to drink from these large flagons, the high incidence of flagons in general at Usk could just mean that they needed more large vessels in which to store liquid, prior to pouring and consuming.
As indicated above, samian and Lyon ware cups were present at Usk and it could be that the lack of Usk ware cups compared to flagons in Usk ware resulted from this importation: there was no need for cups because there were enough samian and Lyon ware vessels. This preference for imported cups could also indicate a difference in status between the imports and the Usk ware cups. In general, the Usk ware cups tend to be smaller than the imported vessels (the mean for the Usk ware cups is 176.66ml, compared to 238.2ml for the Drag. 27 and 202.9ml for the Lyon ware cups). Use of the larger, imported cups as opposed to the smaller, locally made cups could be attributed to a status distinction between officers and soldiers.
Looking at the individual capacities of these groups of vessels can only tell us a limited amount about consumption practices, and the patterns identified are very specific to each group of vessels. To see how the vessel capacities of the different datasets might relate to each other and whether or not there are any significant patterns in the volumes of liquid that they can hold, these data will now be compared across two graphs (Figures 9 and 10).
|Cups||Beakers||Flagons||Other silverware||Total no. of vessels|
There are 67 vessels in total that have had a capacity calculated for this study. I have divided the vessels and vessel form types into groups based on their capacities (Figures 9 and 10, Tables 8 and 9). These vessels and vessel forms need to be compared to see whether or not there are any patterns in the spread of the vessel capacities. There are 57 vessels with a capacity of less than 1 litre. The majority of these are in the 130ml-400ml category, with considerably fewer vessels in the higher and lower categories. However, this is due to the high number of Drag. 27 cups, and the number of larger vessel forms that are represented by only one vessel.
|Jugs||Beakers||Flagons||Total no. of vessels|
|4 litres ->||0||0||3||3|
The group of vessels with capacities of 500ml-1L is distinctive in that it commences at more than 100ml capacity over the remaining vessels with less than 1 litre capacity, and is composed almost exclusively of Menander silverware (Figure 9). The six silverware vessels (M1, M2, M3, M4, M7 and M8) have by far the largest capacity of the vessels in the dataset, and are quite noticeably larger than the next largest silverware vessel (M12). These six vessels are similar in size to the only pottery vessel in this size group (U14). These capacities indicate that, in some situations, the silverware vessels were used to facilitate the consumption of larger volumes of liquid than the pottery and glass vessels.
This group comprises the bulk of the vessels analysed (50 of the 67), 40 of which I have classified as cups. The pottery vessels tend to occur at the low and middle part of this range, i.e. below 300ml, whereas the silverware and glassware tend to be slightly larger, at 200ml and above. The 200ml to 300ml range has the highest density of cups, which might indicate that this is the ideal range of capacity for cups. This, in turn, indicates that the ideal individual portion of liquid was in that range. However, these results are somewhat skewed by the nature of the dataset, notably the numerous Drag. 27 cups included in the analyses. Thus, this is a pattern that is applicable only to this particular dataset. This is especially so since the two largest Drag. 27 cups (nos 55 and 104) have noticeably larger capacities than the rest of the pottery cups, and if there were more of them this could change the pattern to indicate an ideal fill that is higher in volume, or possibly a separate group of vessels with a different ideal capacity.
There are no jugs in this lowest capacity range, and only one Usk ware flagon. The beakers show a great deal of variability, and their spread of capacities show no particular pattern. This can also be seen when the entirety of the beaker assemblage is examined (Figure 9 and Figure 10). The forms that the 'beakers' take are quite varied, and the term 'beaker' covers a wide range of forms (see Figure 7 for the Lyon ware beakers L20.1, L20.4, L20.5, and Figure 4 for the Silverware beakers M25, M26).
It is unlikely that there is one universal function in this disparate dataset of these smaller vessels. The Lyon ware beakers in the 130ml-400ml range are larger than the Lyon ware cups, while the glassware beaker (no. 426) is the smallest of the glassware vessels. The similarities in capacity between the beakers and the cups could indicate that they were used for similar beverages but were used in a different fashion or in a different setting.
This category has ten different vessel form types, but each form type is only represented by one vessel. The ten vessel forms in this category are the largest examined here, and comprise two distinct groups: forms capable of holding between 1 and 3 litres, and those capable of holding over 4 litres (Figure 10). There are no cups, and the majority of the forms are Usk ware flagons.
The group of form types capable of holding over 4 litres have capacities that suggest they were used to pour into other vessels. The three Usk ware flagons (U1, U2 and U5) that hold over 4 litres are noticeably larger than the other forms and are unlikely to have been used for direct consumption.
Flagons U4 and U7 are also more suitable for pouring than for drinking from directly. This indicates that a large volume of liquid could be poured from such flagons, and while several of these vessels are larger than 2 litres, U4 and U7 have a similar capacity at almost 2 litres. A vessel that held roughly 2 litres of liquid could fill up to eight of the Drag. 27 cups: the same as the number of men in a contubernium, although this is likely coincidental because we do not know how these vessels were used or how the different forms would be used interactively. That U4 and U7 are suitable for pouring from indicates that these forms were suitable for individual consumption.
There are two exceptions to this style of individual consumption: the two-handed Usk ware flagon (U8) and the Lyon ware beaker (L20.5), both of which could be used to drink from directly. These two vessels are also similar in size: 2200ml for the flagon, and 1850ml for the beaker. Because of the large volume that could be held in these forms it seems unlikely that these were intended for a single person to drink from. If they were indeed used to drink from directly this suggests that one vessel could be shared between several people, thus making them suitable for use in communal consumption.
The pottery remains from Usk and the comparisons with the vessels of different materials from other sites present a number of possibilities concerning liquid consumption. Most of the vessels of a size for individual use tend to hold less than 350ml, with Drag. 27 cups tending towards 210ml-260ml and Lyon ware cups holding slightly less at 170ml-220ml. The Usk ware cups seem to have a similar capacity range to the Lyon ware cups, but it is important to bear in mind that the Usk ware cups comprise a very small sample size (three vessels – U23, U24 and U25). However, the pottery vessel capacities do indicate that individual portions of liquid at Usk were quite modest.
There are few large vessel forms that can be associated with drinking, and seven out of the eight large vessel form types examined from Usk were Usk ware vessels. While this bias could be a peculiarity of survival, it could be that these large vessels were made to fill a shortfall. The Lyon ware beaker has a similar capacity to the two-handed Usk ware flagon, and both are suitable for direct consumption. The two largest flagon form types (U1 and U2) can also be used to drink from directly, even if this was not their primary function. So, of the large vessels from Usk, at least a quarter of the form types, or as many as a half, could have been consumed from directly and could have held such a volume that they appear to be communal vessels. The two-handed flagon form type (U8) makes up a higher percentage of the finds than the other flagons or cups, which indicates that, while the soldiers at Usk were being supplied with vessels for individual use, and to a certain extent making individual cups and beakers in Usk ware, there was a desire for drinking vessels that had a larger capacity. That these vessels could be consumed from directly indicates that the soldiers preferred to have the option of vessels that could be consumed from communally, not just vessels that were for individual use.
These capacity data also allow opportunities to postulate what was being consumed at Usk. Silverware tends to be associated with wine consumption (Dunbabin 2003, 65-67), and there is evidence for beer consumption in Vindolanda (Vindoalnda Tablets, 182, 186), but there is no way of knowing what this beer was consumed from. If the silverware capacities are taken to be indicative of the volumes used in wine consumption in Pompeii, then how these capacities compare to those of the other vessels in this study may indicate what was being consumed in Usk. The largest silverware vessels (M1, M2, M3, M4, M7, and M8) are only similar to one other vessel (U14), while the rest of the silverware has roughly the same capacity as the other vessels (Figure 9). The similarities in capacity between wine-drinking silver and glass vessels and the pottery at Usk indicate that if beer was being consumed, then it was served in vessels that may also have been used for wine.
The examination of the capacities and forms of these vessels allow some possibilities for identifying their use, but it does not tell us exactly what was consumed. After all, there is no direct evidence for beer consumption at Usk as there is at Vindolanda, and there is no reason to assume that the vessels were used in one specific way all the time.
Specific measurements of volume are mentioned in ancient sources; there are modii and sextarii in the Vindolanda Tablets, and sextarii of wine as part of the wine ration in a fourth-century CE receipt (Roth 1999, 39-40). The volumes of liquid that constitute a similar range to the vessel capacities discussed here are shown in Table 10. These volumes have been included to see if these measurements bear any resemblance to the capacities calculated here, and if they could indicate what an ideal volume would have been for the vessels.
If the measurements were used as a guide, the vessel capacities would be expected to cluster around these volumes. Instead, there is no indication from these calculated vessel capacities that this was the case. The majority of the vessels have a capacity that lies between the volume for a quartarius and a hemina, but do not seem to belong to either measure. If a lack of precision is assumed in the measurement, then it is possible that the Drag. 27 vessels belong to the hemina category. However, for this to be the case, the measure would have to allow for up to 25% variation, which does not seem realistic if these measurements were the target of production. Therefore, such measures of volume have no bearing on the capacities of these drinking vessels, which raises the questions about what the ideal volume would have been, and if there were specifically named volumes for these vessels.
This study is not intended to provide absolute patterns of use for Usk, but rather to explore the possibilities of analysing vessel capacity across different vessel materials. Focusing on capacities alone does not tell us much more than just that figure, but it does provide an insight into drinking habits that measuring rim size or height does not. However, there are limitations to this study of vessel capacity. Using the line drawings from archaeological reports and manually calculating the capacity is a very time-consuming process that limits the number of vessels that can be analysed. The classification of vessels can also be problematic: a system that is applicable to one vessel material may not work for another, and identifying liquid consumption as opposed to food consumption can also be difficult. It is possible that the vessels were made with an end use in mind, but actual vessel use was likely to have been situational – i.e. a vessel that had been intended for wine consumption could also have been used for beer, water or foodstuffs.
Dividing vessels into four classifications based on mathematical ratios alone is too simplistic an approach that fails to account for differences in the capacities, size and also the possible uses of the vessels, as can be seen with the variation within the 'beaker' form and the range of possible uses and sizes of the 'flagon' form. The naming of vessels also brings overtones that influence the interpretation of vessel use. My interpretation of flagons as drinking vessels, or at least as possible drinking vessels, is based on an initial association of the word 'flagon' with liquid consumption. That the flagons investigated in this article have large capacities which would make drinking from them quite difficult does not diminish this association with drinking. Approaches that take this into account by using more neutral names or designations would help to eliminate this bias.
Calculating vessel capacity allows another level of comparison between vessel forms and vessel materials, beyond the measurements of height, rim size etc. However, the classification criteria of the vessels would have to be standardised before this would be feasible for larger datasets. The classification of the vessels, and in particular the comparison between vessel materials, has been problematic throughout the writing of this article, and this is when fewer than 70 vessels have been studied. For a larger dataset, calculating the capacity of each vessel individually, as has been done here, is not practical. Automation of the scanning and calculation (see Christmas and Pitts, this volume), could allow many more vessels to be examined in a considerably shorter length of time.
Considering how varied the capacities of these vessels are, it seems that the soldiers at Usk either did not mind, or did not notice, what volume of liquid was in the vessel, or it may be that the appearance of the vessels mattered as much, if not more, than the volume. There is also the possibility that a distinction between the imported cups and the Usk ware cups indicates a way of differentiating status between officers and soldiers.
The vessels analysed in this study do not conform to a pattern of volume size that is similar to the list of Roman liquid measurements, and can only be classified by ranges, not specific recorded volumes. Whether this is because the vessels were simply not made to reflect these measurements, or because they were made with other specifications in mind, is hard to say. However, the variability among the Drag. 27 cups would indicate that a specific capacity was not important. Whether this only applied to drinking vessels or is also true of other types of vessel is unknown. Handmade vessels will of course vary, and trying to ascribe a particular 'ideal' volume to them could be a modern preoccupation with exact measurements.
Internet Archaeology is an open access journal based in the Department of Archaeology, University of York. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.
Internet Archaeology content is preserved for the long term with the Archaeology Data Service. Help sustain and support open access publication by donating to our Open Access Archaeology Fund.