Skip to main content

A Possible Amphora Second at Sikyon, Greece


This article examines the role of quality control during the manufacturing process of Roman pottery. The criteria used by ancient potters to determine whether a finished vessel was suitable for sale and use or instead should be discarded as a waster has seen limited attention. Additional focus on this topic provides a means of studying behaviors associated with ancient pottery production and decision-making behind different steps of the process. Of particular interest for addressing the topic is a locally manufactured amphora recovered at the site of Sikyon, Greece, from a late fourth- to early fifth-century CE destruction deposit. This amphora was recovered from a use context but has several evident production defects and an X-shaped graffito incised on one of the handles, and it may have been designated as a second in antiquity. Along with a detailed analysis of the amphora, the discussion examines available archaeological, epigraphic, and paleographic evidence for quality control in Roman pottery production and attempts to shed additional light on potential ancient practices of designating vessels as seconds.1


The analysis of pottery collected from archaeological sites in the Graeco-Roman world sheds invaluable light on chronology, economic connections, and activities at a site. One aspect of ancient pottery that remains understudied and underappreciated, however, is the role of quality control in the manufacturing process. What criteria did potters use to determine whether a finished vessel was suitable for sale and use or instead should be discarded as a waster? This topic has seen only superficial treatment by archaeologists.2 For Roman pottery, Peña has noted the lack of emphasis on topics like wasters and rate of production defects despite their potential for illuminating manufacturing traditions.3 Limited attention has also been given to ancient decision-making where particular pottery vessels were designated as seconds. A “second” is a vessel that was blemished or damaged during production, and thus was of inferior quality but still usable, and was still sold, perhaps at a lower cost. A focus on quality control provides an additional means of studying behaviors associated with ancient pottery production and the choices to be made at different stages of the process.

Recent excavations at the site of Sikyon, located in the northeast Peloponnese region of Greece approximately 15 km west of Corinth, have revealed significant evidence of pottery production during the Early and Late Roman periods that may provide new information around decisions related to quality control.4 Of particular interest for the current study is a locally manufactured amphora characterized by several production defects and a distinct X-shaped graffito incised postfiring on one of the handles; it was recovered from a destruction deposit datable to the late fourth or early fifth century CE. The destruction deposit, which is located approximately 35 m north of the kiln in which the amphora was fired, comprised the remains of a room that may have been engaged in the sale and consumption of wine when the damage occurred. Five additional local amphoras of the same type were recovered from the deposit, none of which bear graffiti or significant production defects, along with imported amphoras and table wares, locally manufactured cooking and coarse wares, glass vessels, metal implements, marble table tops, and nearly five dozen coins. Following an overview of recent research into quality control and production defects in Roman pottery studies, this article aims to provide a detailed analysis of the amphora from Sikyon with the graffito and production defects and consider its potential for shedding additional light on ancient practices of designating vessels as seconds.

Production Defects and Roman Pottery

Since the 2007 publication of Peña’s book on Roman pottery, some scholars have attempted to address the lacuna around production defects and quality control in Roman pottery research. For the site of Sagalassos, located in southwestern Turkey approximately 90 km north of the coastal site of Antalya, Murphy and Poblome have focused on a large potters’ quarter in the eastern part of the site that was operational from the first through the seventh century CE.5 The main products manufactured in these workshops were different shapes of Sagalassos Red Slip Ware. Analysis of wasters found in production dumps in the potters’ quarter provided a way to assess whether particular shapes had a greater likelihood of suffering specific types of defects.6 There were also indications that certain firing practices, such as stacking methods in the kiln, impacted pottery vessels in particular ways, in some cases contributing to the occurrence of defects.7 Wasters provided additional information about the evolution of different shapes and the introduction during later phases of the site of attributes like thicker walls and larger vessel sizes, which may have been a response to manufacturing challenges in earlier periods.8 Murphy and Poblome commented as well on the potential of comparing waster debris with pottery discarded after use in household and other urban contexts. Such comparisons could provide insight into ancient considerations of production defects and the specific issues that led to pottery being discarded as part of the manufacturing process versus entering the marketplace.9

Comparable studies are rare, but recent work on grayware manufacture at the Medieval-period site of Cabrera d’Anoia, located approximately 40 km to the west of Barcelona in Spain, also focused on quality control as a central research question.10 The research team was able to determine through use of x-ray diffraction that vessels sent to market had been fired within a specific temperature range of 800–850°C. Vessels fired outside that range were much more likely to bear some type of defect, leading to their discard as wasters.

Focus on production defects has also extended to the potential marketability of vessels that bore some type of blemish in antiquity. Was it common practice, for instance, to sell vessels “as is” as seconds, perhaps at a lower price? Tol and Borgers, who analyzed pottery production at sites in the lower Pontine plain of west-central Italy, argue in favor of this custom. Employing the somewhat paradoxical term “usable wasters” to define products that bore defects but could still fulfil their intended use, they observe: “A situation can therefore be envisaged in which products with minor defects were still considered tradable and were kept apart to be sold at discounted prices, thereby minimizing the loss of revenue.”11 Because this project consisted of surface survey, however, many of their observations were based on the incomplete vessel fragments that could be collected from the surface.12 This poses an additional difficulty, since a comprehensive assessment of the usability of an entire vessel based on a single fragment must be undertaken with caution. With respect to potential marketing of seconds as an economic practice in the Roman world, direct evidence in the archaeological record is limited. At the site of Deva (modern Chester) in England, Bulmer recorded production defects on Gallic Sigillata vessels recovered during excavations from 1974 to 1978.13 She concluded that 26 or 27 of the approximately 160 vessels bore some type of small blemish (e.g., a warped foot, finger marks on slip, overfiring, crazing). It is not known, however, whether such details were considered defects by either producers or consumers.

Perhaps the most intriguing evidence for the designation and marketing of pottery as seconds has been recorded at the site of Novaesium (modern Neuss), located on the right bank of the Rhine River opposite the city of Düsseldorf.14 Of particular interest is a corpus of Gallic Sigillata finds from the site dated to the first century CE, many of which bear graffiti of some kind. In a detailed study of this assemblage, Kütter noted that approximately one-fifth of the vessels bear an X-shaped graffito on the underside of the base.15 The X is always scratched into the surface after firing and is larger than other graffiti that appear on some of the vessels. Previous interpretations had suggested that these graffiti represented some type of owner or user’s mark, or that they represented a number of some kind (e.g., measure or weight). Kütter argued against these interpretations by noting that a simple X on multiple vessels is not a clear distinguishing mark and that the sheer number of examples across different shapes did not suggest use of the symbol as a number. Examination of the vessels bearing an X graffito instead showed that each one bears one or more production defects.16 This could indicate a form of quality control in which these dishes were considered of inferior quality but still serviceable. Kütter compares this with modern porcelain manufacture, where similar marks on the underside of a vessel have been used to designate a lower-quality product.17 While use of graffiti to designate lower-quality table ware is not widely documented from other sites in the Graeco-Roman world, further study could introduce additional examples.18

The above studies have primarily focused on table wares and, in the case of Cabrera d’Anoia, cooking wares. Amphoras, on the other hand, have seen very little consideration of production defects and the potential for containers to be sold or used as seconds. Since the primary use of amphoras was for packaging of liquid goods (wine, olive oil, garum), in many cases for transport over long distances, quality control thresholds were likely quite high. An amphora that leaked or was not easily stackable in a ship’s cargo hold due to issues with its shape would have been a prime candidate for breakage and loss. It is also possible that marketing was a relevant variable. Would consumers be less likely to purchase a commodity packaged in an amphora with noticeable defects? This could be analogous to modern consumers avoiding canned goods with dents or other blemishes.

Quality control during manufacture of amphoras is well attested in papyrus texts found in Egypt that preserve contracts for production of these vessels.19 According to Mees, these stipulations emphasize the importance of each finished amphora being usable.20 Three mid third-century CE contracts recovered at Oxyrhynchus, P.Oxy 50.3595–3597,21 point to an interest in quality control, with their inclusion of the phrase καλω̑ς ὠπτηµένα (well-fired) when describing the types and quantity of jars expected.22 They also make specific reference to excluding amphoras at delivery with some type of production defect: χωρὶς θεραπευσίνων καὶ <ἐ>πισινω̑ν (without any [amphoras] that have been repaired or are damaged).23 An earlier text, BGU 4.1143.16–1724 (19–18 BCE), which discusses vessels not delivered on time based on a previous contract, along with penalties if delivery does not occur by a specified date, emphasizes that each amphora must κε[καυµ]ένα τῃ̑ καθηκούσῃ ὀπτήσι (be fired in proper heat). In P.Tebt 2.342.2525 (final quarter of the second century CE), the potters are required to deliver 2,000 κου̑φα ἀρεστά (acceptable empty jars).26 Several later texts have the same phrase, with the spelling slightly modified to κου̑φα εὐάρεστα.27

Several contracts preserved in papyrus texts also record a practice of ringing amphoras to assess their quality. Presumably, this would help ensure that the containers had been properly fired and met the quality threshold expected by the buyers. One of these texts, P.Oxy 14.1631.16, datable to 280 CE, has the following clause: ποι[η]σόµεθα τὴν τω̑ν χωρούντων εἰς τὸν οἰ̑νον κ[ο]ύφων κοµπασίαν (we will undertake the ringing of jars to be used for wine).28 Another text from the sixth century CE, PSI 8.953.3,29 even references an individual known as a κοµπαστ(ῃ̑), which may translate as “a ringer [of amphoras].” The process of ringing pottery vessels to assess that they are well fired and well manufactured is also described in the Geoponika (6.3.2), a 10th-century agricultural text that is a compilation of excerpts from earlier farming manuals.30

One reason that papyrus texts found in Egypt preserving contracts for production of amphoras show concern for quality of manufacture may be the types of clay that were available to potters. Ballet et al. note, for instance, that certain types of Egyptian clays, such as those available throughout the Nile Valley and in the northwestern part of the Nile Delta, were of inferior quality, perhaps leading to a higher likelihood of wasters during the production process.31 They point to significant numbers of amphora wasters at kiln sites in Middle Egypt, for instance, to highlight this potential phenomenon.32 While this may be a relevant variable for stipulations in Egyptian contracts, the primary use of amphoras as containers for liquid goods across the Roman world means that concern for quality control was likely high overall. Merchants and consumers would want assurance that an amphora being transported, in some cases over long distances, was not likely to break and cause the loss of product and income.

Some evidence comparable to the X-shaped marks on table ware vessels at Novaesium can be found on amphoras in different parts of the Mediterranean. An example is an amphora handle recovered from the Vougioukalakis plot at Eleutherna in western Crete.33 The plot included remains from a housing unit with two phases of activity, one dating to the fourth quarter of the second century BCE through the mid first century CE and the second from the first half of the first through the early second century CE. The handle, thought to be from a local or regional amphora of Hellenistic date, was found during initial cleaning at the site prior to excavation. On the top surface is a faintly inscribed X. The handle from Eleutherna does not, by itself, demonstrate any production defects, but it is only partially preserved and no other fragments of the amphora were identified. It cannot be determined if the X designated the vessel as a second or served a different purpose, and the handle is broken off right next to the X so it is also impossible to conclude if additional letters or signs were originally preserved.

Similar uncertainty is apparent for two fragments of Graeco-Italic amphoras documented at Cosa, located on the Tyrrhenian coast in northwest Italy. One is among the earliest amphora finds at the site and comprises part of the handle and the upper attachment, recovered from a house in the West Block.34 An X graffito is incised on the top of the handle. Analogous to the Eleutherna amphora, there is no indication on the fragment of any production defects, and too little of the vessel is preserved to assess the purpose of the X. The second fragment was recovered from the area of the Capitolium and includes the rim, parts of the neck and shoulder, and the handle attachments.35 On the neck, a graffito in the form of an anchor is preserved, with an X just above it. An anchor is a common device associated with SES stamps found on Dressel 1 amphoras at Cosa, designating them as products of the Sestius workshop.36 Again, no production defects are mentioned in the catalogue entry or are apparent on the preserved fragments.

X graffiti have also been identified on amphoras recovered as part of shipwreck cargoes. If a designation for a vessel with evident production defects, this could indicate such jars, on occasion, were deemed suitable for travel. Issues with preservation, however, make this evidence difficult to interpret. One example is an Africana IIa amphora found with the Giglio Porto shipwreck, located off the coast of northwest Italy and dated to the first quarter of the third century CE.37 The spike and lower part of the body are preserved, and there is an X graffito on the lowest part of the spike. In the catalogue description of the amphora, the phrase superfici sfaldate, “flaking surface,” appears, which is not mentioned for any other jar discussed in the publication. This is a production defect also noted for the Sikyon amphora considered below. One other Africana IIa amphora from the Giglio Porto assemblage, which was preserved intact, mentions a bulge on the body due to the firing process, but no graffito is present.38 Interestingly, an Africana II amphora found at Cosa also has faint traces of an X graffito near the bottom of the spike.39 Similar to the Giglio Porto container, only the toe and a small part of the lower body are preserved. For the X graffito on the Giglio Porto amphora, Gibbins argues against it being an identification or consignment marker due to its placement in a location that was difficult to see, but he does not offer an alternate explanation.40

Several Africana IIa and North African Dressel 30 amphoras from the Monaco A shipwreck also bear X-shaped graffiti. This wreck dates to the first quarter of the third century CE and is located in the southwest part of the harbor of Monaco.41 Among the relevant finds are an Africana IIa jar fragment comprising the rim, neck, and one handle, with an X incised on the handle and small stamp on the neck (SV inside a rect­angle).42 Mouchot notes that similar X graffiti on handles are found on a few other Africana IIa containers from the wreck.43 One Dressel 30 amphora, which has the rim, neck, and part of the shoulder preserved, has an X graffito on the underside of one handle next to the attachment to the neck.44 The neck of this amphora has a second graffito preserving the letters ICT, perhaps representing a tria nomina, with an X below the I. A second Dressel 30 amphora, also preserved from the rim to the upper shoulder, has a large X graffito on the shoulder. The catalogue entries for all of these vessels do not describe any production defects but each vessel is only partially preserved. Mouchot argues that the graffiti were some type of identification marker, an interpretation with which Gibbins agreed.45

Repair of amphoras that suffered some type of production defect or damage during use is also not commonly attested or documented in the Roman world. Peña discusses pottery repair in detail but was only able to uncover a single example of this process for amphoras.46 The container in question is of Late Roman Amphora 1 type, recovered during excavation of the seventh-century CE Yassi Ada B shipwreck, which is located off the coast of Turkey near the site of Bodrum.47 One of the amphora’s handles had broken off the vessel, damaging part of the rim also. According to van Alfen, who discusses this amphora, there is evidence of the damaged section being smoothed and of the hole caused by the loss of the handle being patched. It is unlikely this damage occurred during firing. Instead, it was probably the result of an accident at some point during transport. At the Oplontis B peristyle building along the coast of the Bay of Naples, Italy, Muslin has documented repair to three Dressel 2–4 amphoras being prepared for reuse as packaging containers.48 All three had tap holes drilled into the lower part of their body that were repaired by placing a sherd into the hole and coating the area with pitch on the exterior. Limited evidence for repairs likely relates to the standard use of amphoras for packaging liquid goods and the need to ensure that they were fully sound, not patched up, so they could travel long distances without loss of their contents.

Excavations at Sikyon

A locally manufactured amphora found during excavations at Sikyon in 2016, with additional fragments recovered in 2017, may provide evidence for the designation of these containers as seconds when they were deemed to be of inferior quality but were still usable. This amphora (C 2017-161) was recovered from a destruction deposit datable to the late fourth or early fifth century CE, found in a structure in the southeastern part of the agora that functioned as an industrial complex during the Roman period (fig. 1). Sikyon has been subject to intensive investigation over the past three decades under the direction of Yannis Lolos, beginning with an extensive survey of the region around Sikyon from 2000 to 2002 that documented close to 250 sites, including 148 that were interpreted as habitation sites.49 This was followed from 2004 to 2008 by an intensive survey of the urban area of Sikyon that included geophysical prospection of much of the site.50 The results from the intensive survey and geophysical work laid the foundation for future excavations.

Fig. 1. 
Fig. 1. 

Plan of the site of Sikyon showing Hellenistic to Late Roman structures (drawing by V. Makresia; courtesy Sikyon Excavation Project).

In 2013, excavation began in the southeastern part of the agora at Sikyon.51 Results of the geophysical investigation had revealed a large structure south of an east-west Hellenistic stoa located in this area. Digging took place from 2013 to 2017 and revealed a Roman-period industrial complex that included pottery kilns, pressing floors, cisterns, and a furnace installation (fig. 2).52 Several phases of activity were documented, including the late first through mid third century CE, the mid fourth through early fifth century CE, and the mid sixth through early seventh century CE. Pottery production is attested during all three phases. During the first phase, at least five kilns with external diameters ranging from 1.6 to 3.8 m were in use in the southern part of the complex that contributed to the manufacture of thin-walled ware, various coarse ware and cooking ware shapes, and at least one amphora type. The kilns are all examples of Hasaki’s Type Ia, circular kilns with central pillars, that represent the most common kiln type attested in Greece from the Bronze Age through Roman period.53

Fig. 2. 
Fig. 2. 

Aerial view of the industrial complex in the southeastern corner of the agora at Sikyon (Y. Lolos; courtesy Sikyon Excavation Project).

During the second phase, only the easternmost kiln appears to have been in operation (fig. 3), while the others were buried in a thick, ceramic-laden fill. The main products fired in the kiln at this time were two types of amphoras, designated as Sikyon C and Sikyon D, following a typology initially established during the intensive survey, and several cooking and coarse ware shapes.54 This phase of activity also corresponds with the pressing floors revealed in this complex and the destruction deposit in which amphora C 2017-161 was found. Following another period of disuse, the easternmost kiln was employed for lime production starting around the mid sixth century. In the early seventh century, a small kiln was appended to the western side of this pottery/lime kiln and constructed from reused materials collected from the surrounding area. It was used for the manufacture of stewpots, pitchers, and lids.

Fig. 3. 
Fig. 3. 

View of the industrial complex at Sikyon looking northwest, with the easternmost kiln in the foreground and the room with the destruction deposit indicated by the arrow (Y. Lolos; courtesy Sikyon Excavation Project).

The destruction deposit in the northern part of the complex was localized to a single room and consisted of a thick (0.45 m) layer of soil, artifacts, and debris directly on top of a plaster surface (fig. 4). At the time of the destruction, this room may have been used for the sale and consumption of wine. Numerous artifacts were recovered, including several African Red Slip dishes, a pithos, a stewpot, 11 complete or nearly complete amphoras, amphora stoppers, funnels, pitchers and table amphoras, glass vessels, several marble table tops, an iron and wood padlock attached to an iron chain, some bronze implements, and 55 coins, most of which date to the end of the reign of Constantius II (r. 337–361 CE). Many of the finds help establish a date for the destruction of the room in the late fourth or early fifth century, which will hopefully be refined with further study (table 1). Work on the final publication of the industrial complex in the southeastern corner of the agora at Sikyon is under way, including a more detailed analysis of this destruction deposit and its finds. It will provide a fascinating overview of different activities throughout the Roman period in this part of Greece. The mid fourth- to early fifth-century CE phase will be of particular interest for its potential to shed light on interrelated operations that include pottery manufacture, wine production, and sale and consumption generally.

Fig. 4. 
Fig. 4. 

Destruction deposit in the northern part of the industrial complex, where amphora C 2017-161 was found (full scale = 1 m) (M. Maher; courtesy Sikyon Excavation Project).

Table 1. 

Summary of finds from the destruction deposit in the northern part of the industrial complex at Sikyon.

CeramicAfrican Red Slip form 50Blate 4th c. CE2Hayes 2008, 221 no. 995, fig. 32
CeramicAfrican Red Slip form 52Bmid to late 4th c. CE1Hayes 2008, 222 no. 1002, fig. 32
CeramicAfrican Red Slip form 59Blate 4th c. CE1Hayes 2008, 224 no. 1055, fig. 33
CeramicAfrican Red Slip form 60mid to late 4th c. CE1Williams and Zervos 1984, 94, fig. 7.11
Ceramicfunnellate 4th/early 5th c. CE3Marty 1993, 127–28, fig. 10.a
Ceramicpitcherlate 4th/early 5th c. CE1Abadie-Reynal 2007, 232 no. 408, pl. 64
Ceramicpedestal kraterlate 4th/early 5th c. CE1Slane and Sanders 2005, 257 no. 1.42, fig. 3
Ceramicpithosearly 5th c. CE1Albarella et al. 1993, 174 no. 31, fig. 8
Ceramicstewpotlate 4th/early 5th c. CE1Slane and Sanders 2005, 256 no. 1.35, fig. 3; Hammond 2015, 321 no. 313, fig. 35
CeramicSikyon C amphoralate 4th/early 5th c. CE5Slane and Sanders 2005, 255 no. 1.26, fig. 3
CeramicProto-Late Roman Amphora 2second half 4th c. CE2Slane 1990, 109, 117 no. 255, fig. 29; Marty 1993, 127–28, fig. 11
CeramicProto-Agora M325 amphoralate 4th c. CE1Williams and Zervos 1983, 25 no. 67, pl. 10; Slane 1990, 117 no. 259, fig. 29
CeramicLate Roman Amphora 4final quarter 4th c. CE1Majcherek 1995, 166–68, n. 39, pl. 5.1
CeramicAttic-type post-glazing lampsecond half 4th to early 5th c. CE1Koutoussaki 2008, 251–53 no. 353, pls. 40, 41
CeramicAttic-type post-glazing lamplate 4th c. CE1Wohl 2017, 124 no. 319, fig. 2, pl. 39
Glassvarious shapeslate 3rd to early 5th c. CE23identified and dated by Anastassios Antonarasa
Metalcoins of Constantius II351–361 CE25identified by Irene Marathakib
Metalcoins of Julianus355–361 CE3identified by Irene Marathaki
Metalcoins of Constantius II and Julianus355–361 CE4identified by Irene Marathaki
Metalcoins of Constantius II, Gallus, and Julianus355–361 CE2identified by Irene Marathaki

a Glass specialist, Museum of Byzantine Culture, Thessaloniki.

b Numismatic specialist, German Archaeological Institute at Athens.

View Table Image

Sikyon Type C and D Amphoras

Among the assemblage recovered from the destruction deposit were a small number of Sikyon C amphoras manufactured in the southern part of the industrial complex. Production of this type is documented at the site during the latter part of the fourth century and the early part of the fifth century CE. The standard Sikyon C has a raised base that is convex on the underside with a raised button at the center, and an ovoid body with rounded shoulders that lead to a tall, slightly swollen neck. An outward-thickened rim with square lip often has a shallow groove on top. Vertical handles, round in section and often with ridges, attach to the upper shoulder and midpoint of the neck. The handles regularly bear a thumb impression at the lower attachment. Wheel ridging is present across the upper third and bottom third of the vessel, or in some cases across the entire body.

The Sikyon C type was identified during the intensive survey at Sikyon as a potential new local form.55 The recovery of wasters and hundreds of fragments from kiln dump deposits in the southern part of the industrial complex confirmed its manufacture at the site during the Late Roman period. Examples have been recovered at Corinth, as well, including one described as regional in origin.56 Sikyon C amphoras are reminiscent of amphora traditions within the Adriatic region and were suggested to be of western origin when first described at Corinth.57 The morphology is very similar, for example, to the Forlimpopoli and Empoli types, both of which have a flat bottom, ovoid body, tall neck, and outward-thickened rim, and also to flat-bottomed types produced on Sicily such as the MR1.58 Keay 52 amphoras, a common western type that appears by the fourth century, also have comparable traits like a flat bottom.59 A second amphora type manufactured at Sikyon and identified as Sikyon D appears to have imitated the Keay 52 amphora based on its morphology.60 Regional amphora traditions during the fourth and fifth centuries CE around the Gulf of Corinth focused on a range of types with similar morphological attributes, including a flat bottom. This includes the Agora M325 and its predecessors, and several flat-bottomed types for which production is documented at Delphi.61

Amphora C 2017-161 (figs. 5, 6) may be described as follows:

Sikyon catalogue number: C 2017-161

Fabric: 15–20% well-sorted fine to coarse sand inclusions; common subangular to subrounded white inclusions; frequent angular to subangular gray inclusions; few angular to subangular red inclusions; rare fine silver sparkling inclusions; few spherical and elongated angular voids

Color of surface: Munsell Soil 10YR 5/1 (gray)

Color of exterior core: Munsell Soil 2.5YR 5/3 (reddish brown)

Color of interior core: Munsell Soil 10YR 6/8 (light red)

Color of slip: Munsell Soil 10YR 8/3 (very pale brown)

Diameter of rim: 5.5 cm

Height: 56.2 cm

Maximum diameter: 33.6 cm

Diameter of base: 7.2 cm

Weight: 5.94 kg

Fig. 5. 
Fig. 5. 

Sikyon Amphora C 2017-161 (full scale = 10 cm) (drawing by A. Ognier; courtesy Sikyon Excavation Project).

Fig. 6. 
Fig. 6. 

Sikyon Amphora C 2017-161 with production defects noted (full scale = 10 cm) (A. Siatou; courtesy Sikyon Excavation Project).

This amphora was recovered from the industrial complex destruction deposit. It has the standard morphology of a Sikyon C container. An overview of its characteristics does not, at first glance, suggest any overt issues with the amphora. Macroscopic examination of the vessel does reveal, however, that it is overfired, with the surface flaking away in many places (see fig. 6). Spalling is evident in some places, where bigger patches have flaked away, exposing larger inclusions within the clay fabric. There is a greenish hue in some areas that suggests the amphora was nearing the point of vitrification before firing ceased. Some areas of lighter-colored clay on the surface (10YR 8/3, very pale brown) may indicate the presence of a slip, a feature of other Sikyon C amphoras, but much of this has flaked away. Large areas of blackening or sintering are also present across the body. Perhaps of greater concern is the stance of the amphora, which is noticeably askew when it is stood on a flat surface. (The amphora does stand without needing support, but it is quite nerve-wracking to step away from it since it appears ready to tip over at any moment.) It leans in two directions, both to the side and either to the front or back depending on the viewer’s perspective.

An additional unique characteristic of amphora C 2017-161 appears on the top of one of the handles. A large X-shaped graffito was incised into the handle after the vessel had been fired (fig. 7). Comparable graffiti have not been identified on any other amphoras manufactured at Sikyon, and graffiti on pottery at the site is very rare overall. The presence of this graffito on an amphora that bears multiple production defects may indicate that it served to identify the container as being of inferior quality. It was clearly deemed usable, since it was found in a room with an assemblage pointing to sale and consumption of wine. This may have been the extent of its usability, however, and it is unlikely the amphora would have been suitable for transport of wine or another commodity over short or long distances, even just to neighboring sites.

Fig. 7. 
Fig. 7. 

X-shaped graffito scratched postfiring on one of the handles of amphora C 2017-161 (A. Siatou; courtesy Sikyon Excavation Project).

Characteristics such as being overfired and being noticeably askew do distinguish amphora C 2017-161 from other Sikyon C amphoras, particularly the four additional jars that were recovered from the same destruction deposit (fig. 8). Three are complete or nearly complete, and one has a consistent, clean, even break across the entire upper shoulder, possibly because the neck and rim were sawed off to access the wine inside. The standard color of Sikyon C amphoras appears to be in the light red (2.5YR 6/8) to reddish yellow (5YR 6/6) range. A small production defect is evident on one of the other Sikyon C amphoras from the destruction deposit (C 2017-163), a small palm-size indentation on the shoulder that may have been imposed while handling the vessel, placing it in the kiln. No other part of that amphora is affected by defects, and the indentation is easy to miss at first glance. The tallest complete Sikyon C amphora from the destruction deposit (C 2017-165; see fig. 8) is 64 cm while the other two complete examples (C 2017-163, C 2017-164) are 51 and 53 cm, respectively, suggesting that even with a problematic stance, amphora C 2017-161 is within a standard size range.

Fig. 8. 
Fig. 8. 

Comparison of two amphoras from the same deposit: left, C 2017-161; right, C 2017-165 (full scale = 10 cm) (A. Ognier; courtesy Sikyon Excavation Project).

Is This Amphora A Second?

Does the X graffito inscribed on the handle of amphora C 2017-161 mean it was designated as a second in antiquity? It is unlikely that a graffito in the form of an X had a universal meaning across the Roman world. An association with other variables, such as evident production defects on a pottery vessel, is needed to suggest the X referred to lower-quality products. There was a tradition of assigning products to a higher or lower grade in the Roman world. Pliny describes such products in his encyclopedic Historia Naturalis. One example comes from his description of secundarium passum (second-grade raisin wine), where the detritus from wine pressing, such as the grape skins, was added to water.62 Similar grading of products can be seen in Diocletian’s Edict on Maximum Prices dated to 301 CE. For glass products, of the two entries for window glass, the first reads Speclaris optimi libra una (Window glass, best, one pound), and the second reads Secondi libra una (Window glass, second [quality], one pound).63 The higher quality window glass was valued at 8 denarii communes and the lower quality at 6 denarii communes. There is no indication that the different products would be marked in some way, but merchants must have had mechanisms for demonstrating these distinctions to buyers.

As discussed, Kütter’s assessment of X graffiti on fine ware vessels found at Novaesium included critique of potential alternate interpretations, which can be considered also for the Sikyon amphora.64 He argued against these graffiti on the pottery that he examined being identification marks or abbreviations for a name or number. Those interpretations also seem infeasible for amphora C 2017-161. An identification mark is difficult to contextualize given how rare graffiti on pottery is at Sikyon. Lack of comparable evidence also makes assessing this mark as an abbreviation for a name or number challenging. Perhaps an additional possibility is that the X represents a measurement of some kind. In that context, χου̑ς (chous) would be a logical supposition as a measurement commonly used for liquid goods.65 A χου̑ς, however, is approximately 3.1 liters, a quantity far too small to represent the contents of a single amphora. Το be the plural χόες, a preceding symbol indicating an amount would need to be present. Use of X to designate a price is also unlikely. Within the Roman Empire, a common monetary symbol was X, which typically denoted denarius, even in the eastern Mediterranean.66 The strikethrough was a standard part of that symbol and is not evident on the amphora handle. A small χ could represent the Greek denomination of χαλκοί (chalkoi). A single χαλκός, however, was valued at 1/96 of a denarius, which would be an extraordinarily low price for a pottery vessel this size, even one with evident defects.67 Price data for pottery in the Roman world is limited, but Diocletian’s Edict on Maximum Prices does include an entry for a type of vessel that Peña interprets as a standard wine container of the period with a capacity of 24 sextarii (approximately 13 liters).68 The listed price is 12 denarii communes. In many ways, the fact that the X on these various pottery vessels resembles a letter in both the Greek and Latin alphabets can be distracting. A single line scratched on a piece of pottery could easily be mistaken as an accident or unintentional minor damage. Instead inscribing two lines that overlap makes clear that the mark is intentional.

The combination of the X graffito with the evident production defects on amphora C 2017-161 support the identification of this symbol, in this context, as an indicator of lower quality. For table ware, Kütter argued that the X was inscribed prior to specific vessels entering a supply chain, which moved them from producer to consumer.69 It provided the necessary evidence that the vessel was of lower quality and, likely, of lower cost. Amphoras designated as seconds, however, may have been treated differently. As seen with the evidence from papyrus texts discussed earlier, there were clear concerns for quality of production, which was tied to the use of amphoras as packaging containers for liquid goods; the risk of breakage and loss of the contents required that a certain quality threshold be met. For the Sikyon amphora, while it could still serve as a container for liquid goods like wine, there may have been low confidence in its ability to travel. The fact that it was found, in use, only a few meters from its site of production may indicate the X not only identified the amphora as a second but also designated it for only local use in the same facility.


Amphora C 2017-161 from the destruction deposit in the industrial complex at Sikyon has characteristics indicative of production defects. While it was still usable, the X-shaped graffito incised on one of its handles may have served as a mark of a lower-quality vessel, for which function may have been limited. Comparable evidence of deliberate marking of pottery vessels as seconds, or lower-quality products, is patchy and difficult to corroborate. In the Roman economy, evidence does exist for the ranking of products into different tiers or grades, which likely was the case for pottery as well. While research into quality control in pottery manufacture in the Roman world continues to be lacking despite ongoing discovery and investigation of production sites across the Mediterranean, further focus on this topic can provide a means of uncovering important insight into decision-making and behavior in the Roman economy. Amphora C 2017-161 from Sikyon provides one small piece of evidence for a possibly broader phenomenon that awaits more in-depth consideration.


1 I thank Yannis Lolos, Director of the Sikyon Excavation Project, who invited me to participate in the excavations starting in 2013. Dr. Lolos read a draft of this manuscript and provided several helpful comments and revisions. A preliminary version of this paper was presented as part of a colloquium series at Queen’s University in Kingston, Ontario, in February 2023, and I thank Fabio Colivicchi and Jon-Mathieu Carbon for the invitation to present this research. Images used in the paper were provided by Dr. Lolos; Alice Ognier, the project’s illustrator; Amalia Siatou, the lead conservator; and Matthew Maher, trench supervisor who oversaw excavation of the destruction deposit. I also thank the AJA Editors-in-Chief Emma Blake and Robert Schon for their support and guidance bringing the article to publication. The comments and recommendations of the three anonymous reviewers for the AJA helped in addressing several shortcomings in the manuscript.

2 Widely consulted reference manuals, such as those by Rye (1981, 111), Rice (1987, 105–9), and Orton and Hughes (2013, 146), offer little discussion of the topic of quality control beyond brief coverage of pottery firing and the potential for wasters among finished products. Ethnoarchaeological studies of pottery manufacture also have paid limited attention to this topic. Stark (2003), in an overview of ethnoarchaeological studies of pottery manufacture, does not reference production defects as a topic of study, while Peacock (1982, 14, 50) makes brief reference to the proportion of wasters at some production sites included in his analysis. This is despite discussion of such practices in earlier ethnoarchaeological publications. A study of potters in Afghanistan in the 1960s, for example, recorded that finished products that were blemished in some way after firing would be sold as seconds. See Demont and Centlivres 1967, 57; Rye and Evans 1976, 122–23.

3 Peña 2007, 33. One earlier study is an analysis of defects on vessels manufactured in different fabrics at the site of Greenhouse Farm in Cambridgeshire, England. This was a small part of a larger analysis of pottery production at the site in the mid first century CE, in which the authors were able to demonstrate that certain defects occurred more regularly on vessels manufactured in particular fabrics. See Gibson and Lucas 2002, 107–8, fig. 7.

4 For preliminary results of the excavations, see Lolos 2015; 2016; 2017; 2018.

5 Murphy and Poblome 2011; 2017.

6 Murphy and Poblome 2011, 34.

7 Murphy and Poblome 2017, 65–66, 73–76.

8 Murphy and Poblome 2017, 76.

9 Murphy and Poblome 2011, 34–35.

10 Allepuz et al. 2019.

11 Tol and Borgers 2016, 363.

12 Tol and Borgers 2016, 349–53.

13 Bulmer 1980, 87.

14 Kütter 2008.

15 Kütter 2008, 80–99.

16 Kütter 2008, 88, table 22. Production defects that Kütter identifies include imperfections such as an uneven rim, one or more holes, warped body or rim, uneven base, scratches, extra lumps of clay stuck to the vessel.

17 Kütter 2008, 83–86.

18 At the site of Hierapytna in southeast Crete, a base from a grayware plate likely of Late Hellenistic date has a graffito on the underside of the floor comprising an X surrounded by a square made of four detached lines. The vessel is clearly overfired and the graffito was applied after firing. It could be another example of the phenomenon documented by Kütter. See Gallimore 2015, 109 no. 107, fig. 5.5.

19 Gallimore 2010, 172–77.

20 Mees 2004, 238.

21 In Bowman et al. 1983: P.Oxy 50.

22 In Bowman et al. 1983: P.Oxy 50.3595.34; P.Oxy 50.3596.32; P.Oxy 50.3597.32. For a detailed overview and publication of P.Oxy 50.3595, with reference to the other two texts, see Cockle 1981.

23 In Bowman et al. 1983: P.Oxy 50.3595.36; P.Oxy 50.3596.33–34; P.Oxy 50.3597.33–34.

24 In Schubart 1912: BGU 4.

25 In Grenfell et al. 1907: P.Tebt 2.

26 See Mayerson (1997, 47–48, 51) for discussion of “empty jars” as the most appropriate translation for κου̑φα. Jars in the context of these papyrus texts typically refers to amphoras.

27 In Hasitzka et al. 1986: CPR 10.39.10 (443 CE); in Fantoni 1989: CPR 14.2.16 (late sixth/early seventh century CE); in Maspero 1911: P.Cair.Masp. 1.67110.41 (sixth/seventh century CE); in Preisigke and Bilabel 1926: SB 1.4675.1 (sixth/seventh century CE).

28 The same clause is attested in Coles and Haslam 1980: P.Oxy 47.3354.16–17 (257 CE). See also Skeat 1974: P.Lond 7.2049.3 (third century BCE).

29 In Vitelli and Norsa 1927: PSI 8.

30 For discussion of the Geoponika, see Dalby 2011, 9–18.

31 Ballet et al. 1991, 131.

32 Some sites that Ballet et al. 1991 reference are Sheik Abada/Antinoopolis (134–36), Ashmunein/Hermopolis Magna (137–38), Zawyet el-Maïetin (138–39), the area of the temple at Esna (139), and Deir el-Shohada near Esna (139–40).

33 Bowsky and Gavrilaki 2010, 183–85, 221 no. 18, fig. 3.

34 Will and Slane 2019, 8 no. 6, pl. 1.

35 Will and Slane 2019, 11 no. 15, pls. 1, 50.

36 Will 1979, 343 n. 13.

37 De Tommaso et al. 1991, 126 no. 142, fig. 98. For the shipwreck, see Parker 1992, 193 no. 453.

38 De Tommaso et al. 1991, 125 no. 135.

39 Will and Slane 2019, 82 no. 374, pl. 27.

40 Gibbins 2001, 324.

41 Parker 1992, 279–80 no. 708.

42 Mouchot 1969, 170 no. A8, pl. 2.

43 Mouchot 1969, 170.

44 Mouchot 1969, 170–71 no. A41, pl. 8.

45 Mouchot 1969, 171; Gibbins 2001, 324.

46 Peña 2007, 75–76. In contrast, his section on repairs to large storage containers known as dolia or pithoi covers 18 pages (Peña 2007, 210–27).

47 van Alfen 1996, 202.

48 Muslin 2019, 161.

49 Lolos 2011. Lolos also engaged in a preliminary investigation of the landscape and boundaries around Sikyon from 1996 to 1998, prior to conducting the extensive survey.

50 Lolos 2021.

51 Supra n. 4.

52 An additional set of structures was excavated in the northwestern part of the agora between 2014 and 2019.

53 Hasaki 2002, 154–55.

54 For the typology of Sikyon amphoras based on results of the intensive survey, see Trainor 2015, 48–55; Tzavella 2021, 236–43.

55 Tzavella et al. 2014, 93.

56 Slane 2000, 304, fig. 6a; Slane and Sanders 2005, 255 no. 1.26, fig. 4; Hammond 2015, 310 no. 220, fig. 21.

57 Slane (2000, 304, fig. 6a) suggested that this type was an unidentified western form.

58 Panella 2001, 195–96; Franco and Capelli 2014.

59 Keay 1984, 268–78; Pacetti 1998.

60 This type is rarer than the Sikyon C, and no complete examples have been reconstructed. Most documented fragments recovered from the kiln dump deposits, including several wasters, are heavily encrusted owing to later use of this area for lime production. There is sufficient material to conclude that the base and body had a morphology comparable to Sikyon C amphoras: the Sikyon D has a shorter neck that is concave rather than swollen; handles, oval in profile, attach to the mid neck and upper shoulder; the rim is quite distinct, with standard examples being everted and concave on the upper surface, while some have distinct triangular profiles with tapered lips, and others are more rounded and form an exterior flange, which is characteristic of Keay 52 vessels.

61 The standard Agora M325 is very similar in morphology to the Sikyon C and is a closely related type. For predecessors of this type, see Williams and Zervos 1983, 25 no. 67, pl. 10; Slane 1990, 117 no. 259, fig. 29. For the standard type, see Robinson 1959, 115 no. M325, pl. 32; Slane and Sanders 2005, 255 no. 1.27, fig. 3. For production at Delphi, see Pétrides 2010, 47–53.

62 Plin., HN 14.82.

63 For Latin text and translations, see Whitehouse 2004, 189.

64 Kütter 2008, 80–81.

65 For a discussion of χου̑ς as a unit of measurement, see Kruit and Worp 1999, 102–8.

66 Buttrey 1957, 57 n. 1.

67 Howgego 1985, 57.

68 Peña 1999, 36. The relevant entry can be found in section 15.100, based on the layout of the inscription proposed by Giacchero (1974), under the heading “De fictilibus.”

69 Kütter 2008, 82.

Works Cited