Characterizing Roman Artifacts to Investigate Gendered Practices in Contexts Without Sexed Bodies

This article concerns the characterization of Roman artifacts so that they can play a greater role in gendered approaches to Roman sites—sites that constitute lived spaces but lack actual references to sexed bodies. It commences with a brief discussion on gendered approaches in the two main strands of Roman archaeology—classical and provincial. Within the differing frameworks of the wider disciplines of classics and archaeology, both strands focus on contexts with sexed bodies—burials, figurative representation, and inscriptions. The discussion serves as a background for more integrated and more interrogative approaches to relationships between Roman artifacts and gendered practices, approaches that aim to develop interpretative tools for investigating social practice in contexts where no representational or biologically sexed bodies are evident. Three types of artifacts—brooches, glass bottles, and needles—are used to demonstrate how differing degrees of gender associations of artifacts and artifact assemblages can provide insights into gender relationships in settlement contexts. These insights in turn contribute to better understandings of gendered sociospatial practices across the Roman world.


introduction
Hill observed that the "quality of [Roman archaeological] data to address gender issues is considerably greater than for any prehistoric periods, and as good, sometimes better, than much medieval evidence." 1 However, gender, as a sociocultural construct with "constantly negotiated relationships" constituted in historically specific ways, is not inherent in archaeological data. 2 In the geographically and chronologically diverse Roman world, where social status and ethnicity (i.e., slave, freed, free, citizen, peregrine) often played more significant roles in social hierarchies and socioeconomic practices than did biological sex, gender as a defining characteristic of identity and practice is problematic. 3 Categories of material from the Roman world cannot be assumed to have always carried a particular status or gender value without detailed consideration of the assumptions involved. 4 That said, we are often well informed about certain gender associations through textual, epigraphical, and representational evidence and through burial remains found in vastly different regions and periods throughout the Roman world. These types of evidence provide some of the "rules" different groups within that world had about how material culture might be gendered. 5 Thus, feminist and gender archaeology across the Roman world has focused on evidence for gendered identities as represented in these contexts. However, exploration of gendered identities and practices at Roman archaeological sites that lack these types of evidence is limited.
This article is concerned with characterizing Roman artifacts so that remains from lived spaces can be used to greater effect for insights into the presence, roles, and identities of women within these spaces. 6 Certain types of artifacts found during the excavation of Early Roman imperial sites, notably military bases, provide case studies; these studies can be used to demonstrate how material-cultural approaches to the wealth of artifactual evidence from across the Roman world can inform investigations of gendered practices at sites that lack evidence for sexed bodies and where interpretation of the "rules" of social practice has traditionally been rather androcentric. I propose that artifact characterization with differing levels of gendered associations can help decode such material evidence. 7 While I acknowledge the existence of changing and differing gender identities across the Roman world, I argue that any apparent consistencies of gendered practices in artifact use across that world have important ramifications for understanding how sociocultural practices spread. 8

approaches to gender in roman archaeology
The broader disciplines of archaeology and classical studies have quite well-developed bodies of theory and practice in their approaches to women and gender. Roman archaeology has been rather slow to engage with these approaches. Added to this slowness are extant boundaries between approaches within Roman archaeology. Its division into two different, although increasingly converging, 9 strands-classical and provincial-has resulted in different pathways for engagement with the material record and with feminist and gender theory. 10 Past approaches to artifacts, and the labor involved in collecting, analyzing and reanalyzing this wealth of remains, have also been major obstacles inhibiting Roman scholars from developing more theorized, interdisciplinary approaches to interpreting artifacts and gendered practices. The range of approaches, these boundaries, and these disciplinary histories all provide significant challenges for more integrated material-cultural approaches in feminist and gender research across Roman archaeology. 11 In classical Roman archaeology, with its focus on Italy and the center of the Roman world, feminist and gender research is framed by the concerns of its sister disciplines of classical studies and art history. 12 The emphases of feminist Roman social history on elite women's public roles and power relationships, and on their families and households, provide the context for much of classical archaeology's approach to gender. 13 Rather than use archaeological methodologies to analyze material remains, classical Roman archaeology employs essentially art historical approaches to visual representation, to investigate gender perception in the Roman world through the sculpted portraits of real women as well as mythological women in mosaics and wall paintings. 14 Despite close links with the wider classics discipline, Roman archaeological contributions 5 Díaz-Andreu 2005, 23. 6 Women's place continues to be an important line of inquiry for feminist Roman archaeology (e.g., Baker 2003;Revell 2010; see also Spencer-Wood 2006, 301). 7 Díaz-Andreu 2005, 37-9, 42. 8 See Allason-Jones (2011, xiv) on the "core" of artifact types across the Roman world. 9 Eckardt 2010, 7. 10 These two strands and their different approaches have not always been evident to scholars outside Roman archaeology, however (see, e.g., Spencer-Wood 2006;Tomášková 2006, 22). 11 Baker 2003. 12 See, e.g., studies such as Skinner 1987;Rabinowitz 1993;Archer et al. 1994;Ward 1996;Wyke 1998. 13 For public roles, see, e.g., Abbott 1909, 41-99;Gardner 1986, 233-55;Setälä et al. 2002;Dixon 2007;Gregorio Navarro 2013. For families and households, see, e.g., Saller 1984;Rawson 1986Rawson , 1991Rawson , 2011Dixon 1988Dixon , 1992Dettenhofer 1996;Rawson and Weaver 1997;Milnor 2005;Treggiari 2005; see also Hemelrijk 1999;Barrett 2002. Hemelrijk (2012 argues that these concerns are framed by the moralizing approaches of the ancient authors. More theorized gendered approaches are found in Greek social history and archaeology (e.g., Foxhall and Salmon 1998) and in the better-documented Late Roman and Early Christian periods (e.g., Cooper 2007;Osiek 2008) rather than the Republican and Imperial periods. E.g., Foxhall and Neher's (2013) only chapter on the Roman world is on the later Christian empire (i.e., Cooper 2013). Also, historical studies on the complexity of gendered sexuality in Rome have had little impact on Roman archaeology (see, e.g., Hallett and Skinner 1997;Parker 1997;Montserrat 2000;Wyke 2002;Skinner 2005). 14 E.g., Brown 1993;Davies 1997;Harlow 2004; see also Montserrat 2000, 166-75. For real women, see, e.g., Kampen 1981Kleiner andMatheson 1996, 2000;Hemelrijk 2004Hemelrijk , 2008Hemelrijk , 2012Alexandridis 2010;Heyn 2010;Wood 2010. For mythological women, see, e.g., Berg 2010Carucci 2010. Interestingly, my paper on a database of Pompeian house contents read at the 92nd Annual Meeting of the Archaeological Institute of America (San Francisco, 1990) was included in a section called "Roman Art in Context" (Allison 1991). to feminist classical literature have been notably limited compared with those of the Greek world. 15 Indeed, many Roman social historians have been reticent to acknowledge the role that material culture more broadly can play in gendered approaches, arguing that for the classical world "gender issues are not always apparent in the physical record" 16 because "ancient women . . . left behind so few traces of themselves." 17 Such reticence is based on entrenched assumptions of the implicit masculinity of Roman material remains.
In contrast, the discipline of Roman archaeology in peripheral regions of the empire has engaged differently with feminist and gender approaches to material culture. Here it is more closely allied with the wider discipline of archaeology than with classical studies or art history. 18 Feminist and gender archaeology in the northwest provinces in particular is aligned more strongly with Anglo-American and Scandinavian prehistory. 19 A continued focus on women and "remedial and corrective" concerns in the broader archaeological discipline has influenced provincial Roman archaeology. 20 Gendered practices in Roman provincial archaeology have mainly been investigated in funerary contexts, particularly where actual human bodies can be osteologically sexed and have associated burial furnishings that can throw light on gendered artifacts, identities, and practices. 21 A problem for such investigations in prehistory is that they often employ double standards and draw on cultural biases. Roman archaeology's recourse to documentary sources means that such double standards can to some extent be minimized. 22 However, the use of many written sourceswhich tend to present elite male voices from the Roman center-in the less well-documented provinces is often analogical. As Spencer-Wood argued, much scholarship in Roman archaeology, both classical and provincial, can also be considered ungendered in its assumptions about the normative behavior represented by such sources. 23 A good example to demonstrate how artifacts from other parts and periods of the Roman world can indeed be used to document alternative gendering is Cool's analysis of male-sexed skeletal remains wearing jet jewelry from the third-century C.E. Roman-British town of Catterick. 24 Cool identified one of the skeletons as that of a priest of Cybele. While jet jewelry was considered a female attribute and male jewelry wearing was frowned upon in imperial Roman society, men in the provinces, particularly in Iberia, Africa, and the eastern provinces, did wear jewelry, and male dress became more elaborate in the later empire. 25 The ready availability of jet in northern Britain may also have contributed to its different gender significance in this context. This example underscores the pitfalls involved in assuming direct and unproblematic correspondence between artifacts, gender, and status identities, and between burial and lived practices across the Roman world. 26 While feminist and gender studies in Roman classical archaeology in the 1990s focused on elite women and were separated from the more material-cultural approaches in provincial Roman archaeology, this situation is changing. 27 For example, there has been an increasing interest in figurative representations, notably on grave monuments, of differing social groups and of family relationships in peripheral regions of the Roman world. Such studies have concerned the social conditions revealed through these depictions, their expressions of gendered identities and practices, 15 See, e.g., Rabinowitz and Richlin 1993;Cornell and Lomas 1997;Feichtinger and Wöhrle 2002;Skinner 2005;cf. Birk 2010. For Greek archaeological contributions, see, e.g., Kehrberg 1982;Hitchcock 1997;Koloski-Ostrow and Lyons 1997;Osborne 1998;Stafford 1998. It is therefore unsurprising that Spencer- Wood's survey (2006, 296, 315-16, 318-19) of third-wave feminist studies in classical archaeology focused on art historical studies of Greek art. 16 Wallace -Hadrill 1996, 112. 17 Milnor 2005, viii. 18 Trigger 2006; see also Baker 2003, 140. 19 For discussion, see Díaz-Andreu and Stig Sørensen 1997;Díaz-Andreu 2005, 17. 20 Díaz-Andreu 2005; see also Stig Sørensen 2000;Hamilton et al. 2007;Nelson 2007. For Roman archaeology, see Revell 2010. Spencer-Wood (2011 contrasted feminist prehistory with feminist historical archaeology to argue for a dichotomous relationship with a male prehistory and a female historical archaeology. This representation excludes archaeologies of the Greek and Roman world that have long been more pluralist, if essentially less feminist or gendered (see, e.g., Cohen and Sharp Joukowsky 2004 Allason-Jones 1995, 25-6;2012, 473;Matthews 2000, 13;Harlow 2004. 26 Pearce 2010, esp. 84-5; see also Cool 2011, 299-312. For further discussion, see Allason-Jones 1995, 2009 Whitehouse (1998) did not include any Roman studies because, she argued, few of the works from the 1990s are "explicitly archaeological studies" (1). There were no contributions from the classical world in Bacus et al. (1993), let alone from Roman archaeology; see also Zarmati 1994. The only chapters on the Roman world in Moore and Scott (1997) concern analyses of documentary sources (e.g., Harlow 1997). and their symbolic significance. 28 However, much of this research might still be considered empirical and ungendered given that it regards funerary reliefs as akin to photography and subsumes women in terms of the familia. 29 Studies of nonliterary written evidence have played a more significant role in bridging the disciplinary boundaries between Roman social history and archaeology and the two strands of Roman archaeology, and also in broadening feminist and gender approaches to encompass the wider roles of women from different social groups across Roman society. 30 The written voices represented in many inscriptions, however, are often still official voices idealizing social relations. 31 Other types of nonliterary written evidence-graffiti, military diplomas, curse tablets, and the wooden tablets from Vindolanda and Vindonissa-give greater insights into other voices, especially of lower-status women. 32 In summary, while approaches to gender across Roman archaeology are converging, they are still reliant on the sexed bodies as represented in the sources. These sources, with the exception of some nonliterary texts, concern mainly the symbolic gendering of identity and practice. To date, few studies of gender in Roman archaeology have attempted to investigate contexts of actual practice that lack such sexed bodies. Despite Brown's comment two decades ago on the importance of linking "[a]rtifacts such as loomweights and particular kinds of toiletries, clothing, jewelry, vases . . . to patterns of female behavior," 33 and despite general acknowledgement that artifacts associated with sexed bodies inform gendered identities, material-cultural approaches to actual gendered practice are still largely missing from Roman classical archaeology. 34 Spencer-Wood argued that the few studies that have taken critically gendered approaches to artifacts in lived contexts (e.g., in household archaeology) have been ungendered and nonfeminist readings of this material. 35 gendered approaches to roman artifacts and lived spaces Hunter's comment that "artifact research has endured a complicated relationship with the broader field of Roman studies" applies particularly to feminist and gendered approaches to Roman artifacts. 36 Many Roman archaeologists have ostensibly circumvented the concerns of feminist and gender archaeologies by investigating other types of identities, particularly ethnicity and status, and concepts of acculturation and imperialism. 37 As noted above, a major concern for gender archaeology has been the assumed maleness of many Roman archaeological remains. This assumption stems in part from a somewhat circular approach to the producers of mainly architectural or structural remains rather than a consideration for the users of these spaces and for other types of material culture as keys to understanding social practices in these contexts. 38 As also outlined above, the main material sources used by feminist archaeologists to develop insights into the hidden voices across the Roman world are representational, epigraphical, and funerary. This evidence for sexed bodies and for associated material culture has rarely been used, in any systematic manner, to facilitate investigation of gendered practices within lived space. However, such evidence can be interrogated for a more critically gendered characterization of Roman artifacts and of artifact assemblages. These characterizations, in turn, can assist in more gendered approaches to how people throughout the Roman world played out their lives.
Past studies that have explored the gender associations of artifacts in the sexed context discussed above have focused on artifacts associated with dress. 39 Burials that lack osteological analyses to sex the skeletal remains have been gendered according to dress items within the grave assemblages. 40 Böhme-Schönberger's and Martin-Kilcher's examinations of grave assemblages 28 E.g., Boatwright 2005;George 2005;Carroll 2013aCarroll , 2013b. The earlier studies by Kampen (1981Kampen ( , 1982 are exceptions. 29 Swift 2011, 202. For discussion, see Spencer- Wood 2006, 300, 312. 30 See esp. Allason-Jones 1989Revell 2010; D'Ambra 2012; see also Treggiari 1976;Bagnall and Frier 1994;Bagnall 2006. 31 See, e.g., Shumka 2008, 183. 32 Speidel 1996; see also Allason-Jones 2012, 471-72. 33 Brown 1993, 258. 34 For gendered artifacts associated with sexed bodies, see Cool 2011. For more material approaches in Roman classical archaeology, see Berg (2010) on artifacts associated with Venus in Pompeian wall paintings as representations of the mundus muliebris (a coherent group of women's toilet items) and as female attributes of virtue and femininity.  Swift 2011, 203. 40 See, e.g., Effros 2004. For discussion, see, e.g., Gardner 2007, 230. in Germany, Switzerland, and northern Italy have used brooches, jewelry, and other supposed gendered attributes to sex the burials. 41 Such studies might be considered empirical and ungendered readings of these material remains. 42 However, as is discussed below, their combined results demonstrate consistent patterns of gender associations that constitute a useful body of data for understanding gendered practices within these burial contexts and also across these regions; they can inform the interpretation of gendered practices in lived contexts in these regions or in other parts of the Roman world.
Dress-related artifacts have indeed been used to identify gendered practices in some lived contexts that lack sexed bodies. Van Driel-Murray used the size ranges of leather shoes found in Early Imperial military bases to argue for the presence of women and children inside soldiers' barracks. 43 Her studies have instigated a call for a radical revision of our perspectives on Roman military bases as lived space and have set an agenda for more critical and systematic approaches to Roman artifacts and gendered identities and practices in such lived contexts. 44 In these military contexts, in particular, too much emphasis has been placed on documentary sources and structural remains as keys to understanding social identity and practice at the expense of artifactual evidence. 45 Because women and families are largely missing from written evidence on the Roman military life, such sites have been considered hypermasculine, and noncombatant personnel are largely assumed to have been absent from inside the fort walls. 46 However, detailed and systematic analyses of artifacts from such military sites show that this was not the case. 47 Military studies may have been the slowest among Roman studies to confront their "historical patriarchal ideolog[ies]," and Stig Sørensen's criticism of archaeology more broadly for its intellectual baggage applies to the approach of Roman military studies to material remains. 48 Investigations of these military sites, as assumed masculine spaces, have the potential to lead the field in more material-cultural approaches to gendered sociospatial practice in Roman archaeology. Gendered perspectives of military sites, as contexts that essentially lack actual bodies, require more material-cultural approaches than do many other branches of classical studies and much of Roman archaeology. Studies of Roman military life can draw on a wide range of textual, epigraphical, and representational sources that potentially provide the "rules" of social practice. These sources indeed provide evidence for social diversity in this sphere, but they are not generally concerned with the mundane and routine activities of the various nonmilitary members of these communities, including women and children. 49 Such activities are documented by the artifacts left at these and other types of lived sites. 50 I argue that interpretative links can be found between artifacts and gender in contexts with sexed bodies and that such artifact types can be systematically analyzed, characterized, and used critically as tools for investigating gendered identities and practices within archaeological contexts that lack such bodies. 51 gendered characterizations of artifact types and gendered space While van Driel-Murray called for more holistic approaches to artifacts as gender attributes for investigating social identity, a cautious approach is needed to mitigate the risk of stereotyping gender identities and practices across the Roman world. 52 The following discussion demonstrates how more systematic and integrated approaches to all types of evidence, especially from contexts with sexed bodies, can be used to ascribe levels of gender characterization to certain Roman artifact types, such that they and their assemblages can provide insights into gendered sociospatial practices in lived contexts that lack bodies. The examples chosen are specific artifact types whose gender characterizations are by no means precise and assured, and they concern both gendered identities and practices, "being" and "doing" gender. 53 That is, one item of dress, one associated with personal hygiene, and one associated with cloth-working activities are used to demonstrate how interrogative approaches to various types of evidence can ascribe gendered characterizations to artifact types. The examples chosen are all potentially associated with women and are found inside Early Imperial military bases. They demonstrate a range of levels of gender association, from the more probable 41 E.g., Böhme-Schönberger 1985Martin-Kilcher 1993. 42 Spencer-Wood 20062011, 3-4. 43 van Driel-Murray 1994 See, e.g., James 2006, 34;Gardner 2007, 230. 45 For discussion of similar approaches to domestic space, see Allison 2001. 46 On the "ideology of hypermasculinity," see Spencer-Wood 2006, 320. 47 Allison et al. 2005;Allison 2006b, 2006c, 2008a, 2103. 48 Stig Sørensen 2000Spencer-Wood 2006, 321. 49 Phang 20012011, 131-33. 50 Gardner 2007, 200. 51 Stig Sørensen 2006, 28-31. 52 van Driel-Murray 1997, 55. For discussions on stereotyping, see Diaz-Andreu 2005, 17, 27;Gardner 2007, esp. 80, 84, 202, 204, 229-31. 53 Moore 1999; see also Allison 2006b, 5-7. female dress items to artifacts that document activities that tend to be associated with women but by no means are exclusively so.

Artifacts Probably Associated with Women's Dress: Thistle-Shaped Brooches (Distelfibeln)
While there are distinctive and well-known genderand status-related attributes of traditional Roman dress, many scholars have cautioned against the gender stereotyping of various dress-related artifacts. In the western provinces, especially in contexts dating to the Early Empire, the most common dress items found archaeologically are metal brooches. The presence of specific brooch types in military contexts has traditionally been used to argue that these were types worn by Roman soldiers. 54 However, such an argument gives precedence to preconceived assumptions about who occupied these military bases over specific evidence for how different types of brooches would have been worn by different status and gender groups. 55 Brooches were part of both male and female dress in much of pre-Roman Europe and were adopted and adapted during the Roman period. 56 Our understandings of the different ways in which various brooch forms were worn play an important role in ascribing their gender associations. For example, high-bowed brooches were used to fasten coarse and thick material such as overgarments and so were likely to have been worn by both men and women. 57 Flatter brooches were for thinner fabrics, such as in women's undergarments. Roman soldiers also wore certain brooches as insignia, and women wore them as jewelry. 58 While these observations provide general rules, assigning brooch types exclusively to women or men, or to soldiers or civilians, is problematic. 59 Current understandings of how specific brooch types would have been worn, by whom, and in what context have been developed through a combination of detailed typological analyses of brooch forms and how they functioned; analyses of burial assemblages; and analyses of brooches in figural representations. 60 For example, depictions of women on grave monuments indicate that, at least in the German provinces, women's dress required three or more brooches: a pair of high-bowed brooches at the shoulders, a flatter one fastening undergarments, and possibly further decorative brooches as jewelry. 61 A frequently cited example is the grave monument of Blussus and his wife, Menimane, from Mainz-Weisenau. Dated to the Tiberian-Claudian period, the sculpture represents Menimane wearing at least three brooches in this manner. 62 Burial assemblages have played a large part in many gender characterizations of these brooches, although these assemblages are often from burials lacking sexed skeletal remains because they were dug without appropriate analyses of the skeletal evidence. Nevertheless, studies of these numerous and rich grave assemblages conducted over more than 50 years have argued that distinctive assemblages can be used to identify male and female burials. For example, assemblages from the pre-Roman Rhine region and from northern Italy demonstrate that women wore brooches in greater numbers than did men and that this pattern continued into the Roman period. 63 In Schankweiler, near Trier, more than half of the 20 or so graves of the late Augustan to the early Flavian period identified as female had two or three brooches, and four graves had three to six brooches. 64 While different brooch types do not seem sex-specific within the indigenous milieu, these burial assemblages indicate that, by the Augustan period, a distinction had developed such that some types of brooches and ways of wearing them were indicative of status and sex. 65 These distinctions have been used to demonstrate gender, age, and regional identity for later periods in Gaul, in the Danube region, and in Roman Britain. 66 Of significance here is the combination of available representational and funerary evidence that can be interrogated to establish differently gendered dress that, in broad terms, ranges across regions and periods but arose during Roman occupation. My concern is how such sex-specific associations can be used to characterize specific brooch types according to gender so that this characterization can in turn be used to identify gendered behavior and, more specifically, the gendered use of space. A useful example, which also illustrates changing attitudes to brooches as gender attributes in both antiquity and modern scholarship, 54 E.g., Swift 2011, 213. 55 Swift 2011. 56 Böhme-Schönberger 1995. 57 Gechter 1979. 58 Böhme-Schönberger 1995, 4. 59 Allason-Jones 1995, 23-5. 60 E.g., Böhme 1972;Ettlinger 1973;Riha 1979Riha , 1994; see also Böhme-Schönberger 2008, esp. 141 n. 8;Carroll 2013b. 61 Martin-Kilcher 1993; see also Riha 1979, 41. 62 Ludwig 1988Böhme-Schönberger 1995. For further references, see Allison 2013, 71-7. 63 Böhme-Schönberger 2002, 217;2008, 142;Martin-Kilcher 2003, 281. 64 Ludwig 1988, 197. 65 Martin-Kilcher 19931998, esp. 224-27;Böhme-Schönberger 2002, 217;2008, 143. 66 Gaul and Danube : Effros 2004. Roman Britain: Eckardt 2005 is the so-called Distelfibel, or thistle-shaped brooch. 67 This brooch type ( fig. 1), its use, and the contexts in which it has been found have been extensively studied by  She describes the type as a massive, heavy brooch with a ribbed semicircular bow that had a large shield decorated with curved and incised pressed sheet metal. She noted that until the late 1970s these brooches were thought to have been worn by Roman soldiers because they were found inside military forts. 69 However, Gechter observed that Distelfibeln represented less than 5% of brooches found inside military fortifications, while in oppida (i.e., local settlements) double that percentage was found. 70 He argued that this distribution suggests this was a civilian, and quite possibly a distinctively female, fastener.
This brooch type has also been recorded in numerous pre-Roman graves. On the basis of their assemblages, Böhme-Schönberger identified examples of a single Distelfibel in each of only two pre-Roman Late La Tène "male" graves. She identified two or three examples in each of a further four "male" graves in free Germany dated to the Early Imperial period but seemingly outside the Roman milieu. 71 She analyzed reports on Early Imperial burials in France, Germany, Switzerland, Britain, and Denmark and noted that most Distelfibeln occurred in burials with female assemblages. 72 For example, a pair of Distelfibeln was recorded in Roman-period Grave 76 in Schankweiler, which Ludwig identified as a women's grave. 73 Martin-Kilcher also frequently recorded Distelfibeln in association with women's assemblages in Early Imperial burials in the Alpine region. 74 Böhme-Schönberger identified as Distelfibeln the brooches on the overgarments of two women depicted on Tiberian-Claudian grave monuments at Ingelheim am Rhein and the pair Menimane wore at her shoulders in the representation on her grave monument. 75 She therefore argued that Distelfibeln may not have been gender differentiated in their indigenous milieu but that they became typical women's brooches in the Augustan period. 76 She argued that inside the western provinces Distelfibeln were predominantly part of women's dress but continued to be worn by men from free Germany.
The gendered characterization of these thistleshaped brooches is based on grave assemblages, representational evidence, and, in the case of Gechter's study, an assumption of a dichotomous male military space and female civilian space. 77 Nevertheless, the combined weight of all these studies-involving close study of the actual artifacts as well as synthetic analyses of quantity and distribution and covering half a century of data collection and analysis-presents a strong argument that this particular brooch type can indeed be gendered predominantly female in most Romanperiod contexts within the northwest provinces. 78 Consistency can be found in rich Roman-period burial assemblages such that, even without sexed skeletal evidence, it is possible to suggest gender attribution that is supported by figurative representation.
While there are exceptions, there is therefore strong evidence for Distelfibeln as female attributes. This is not to say that this brooch type is a definite female attribute but rather that female is the most prominent gender association for these brooches. Their presence in lived spaces can therefore be used to explore gendered sociospatial practices. While Gechter's quantitative study showed lower percentages of these brooches in military 67 Almgren 1897, no. 240; see also Böhme-Schönberger 1998, pl. 11. 68 E.g., Böhme-Schönberger 1995. 69 Böhme-Schönberger 2008, 140, 143. 70 Gechter 1979. 71 Böhme-Schönberger 2008, 145. 72 For references, see Böhme-Schönberger 2008, esp. 142-44. 73 Ludwig 1988, 197-200. 74 Martin-Kilcher 1993, esp. figs. 5, 7-9. 75 Böhme-Schönberger 1995, 4-5, 9. 76 Böhme-Schönberger 2002;2008, 142-43. 77  contexts than in settlement sites, it is significant that they are not excluded from the former. Provided the taphonomic conditions permit, their presence, associations, and distribution within such contexts can highlight, for example, female participation in particular activities within military bases. Any associated artifacts attributable to women's dress and activities may serve to confirm this and negate an association here with men from free Germany. 79 So a Distelfibel reported from the east gateway of the first-century legionary fortress Vetera I in the lower Rhine region, and the association of this brooch with other potentially female-related items-an Almgren 16 brooch, an unguentarium, two hairpins, and a bone disk that was probably either a bead or a spindlewhorl-can be used as evidence for the association of women with the street life of this fortress. 80 This brooch type may also have been an age and status attribute. 81 Menimane was the wife of a wealthy provincial shipowner who could afford a large Romanstyle grave monument. 82 Many Distelfibeln were found in rich burials, suggesting that this brooch type was a symbol of high status. This characterization might be used to argue that at least one of the women associated with street life inside the Vetera I fortress was not necessarily a tradesperson or prostitute but may have been a high-status local woman. Thus, this brooch and its associated assemblage in this lived context are potentially useful for investigating interactions between local people and the Roman military.

Artifacts Possibly Associated with Women's Toilet Activities: Perfume Bottles
Perceptions of female beauty in the Roman world, derived from written sources and artistic representations, indicate that artifacts associated with personal hygiene, health, and beauty often served as female attributes, especially of elite women. 83 Care of the body and bodily adornment were seen to "soften Roman citizens," and toilet activities served to "display the adorned female body." 84 While toilet items appear to be inappropriate symbolic attributes for Roman men, this does not mean that men did not use them. For example, while mirrors and combs were named as part of a woman's toilet set and were used to symbolize femininity in Roman art, men no doubt used these items. 85 Allason-Jones argued that a nail cleaner and tweezers found in the lived space of the turrets on Hadrian's Wall would have been used by soldiers stationed there. 86 While her argument, like Gechter's, was based on assumptions about who used the space, it warns of the problems of associating all personal hygiene items with female activities. 87 That is, the symbolic association of toilet activities with female beauty does not necessarily represent actual practice. 88 Indeed, many toilet items found in excavations, such as spatulas, probes, and tweezers, could equally have been medical implements and so cannot be easily gendered. 89 Furthermore, the use of the term "medicamentum" for both cosmetics and medicaments points to a lack of differentiation in the Roman world between these substances and, by extension, to a lack of distinction between medical and cosmetic activities. 90 This also applies to equipment associated with these activities.
There is, however, one type of artifact that seems more specifically associated with women's toilet activities, and therefore its presence in lived spaces in excavated Roman sites is potentially significant in terms of gendered practice. This is the small ceramic and glass bottle ( fig. 2), which is widely considered to have been used as a container for cosmetics and perfumed oils. Long, narrow bottles (see fig. 2a-d), frequently called "unguentaria" or "balsamaria" (both terms invented by archaeologists), are found across the Roman world. 91 Squatter and rounder bottles (see fig. 2e), often referred to as aryballoi by modern scholars, are frequently found associated with bathhouses and are used by scholars to reconstruct toilet sets. 92 Examples of these types of bottles, particularly long, narrow ones, 79 Böhme-Schönberger 2008, 145. 80 For the Almgren 16 brooch, see Hanel 1995, cat. no Swift (2011, 207) argued that younger women wore more jewelry than older women. 82  are represented in Roman art as parts of cosmetic sets but are also found in association with medical equipment, concurring with the lack of distinction between cosmetics and medical remedies. 93 Because of their presumed use as perfume and cosmetic bottles, however, these small glass bottles are considered to have been predominantly used by women in their toilet. 94 Perfume and perfume bottles are listed in the Digesta as part of the mundus muliebris. 95 Again, though, an exclusive association of perfumed oils, and therefore perfume bottles, with women is not substantiated. Despite attention to male grooming being considered a vice in Roman society, in Rome elite men used perfumed oils after the bath, and perfumed oils could be used for anointing military regalia and statues of deities. 96 Berg noted that perfume bottles are rarely depicted among the vessels considered symbolic of women's toilet in Pompeian paintings but that the vessels associated with women's bathing in these paintings are predominantly Greek types. 97 This is no doubt because these paintings are usually copies of Greek originals, representing subjects of Greek mythology and probably Greek practice. Indeed, likely perfume bottles are represented in what appear to be mundi muliebri on two second-century women's grave markers from Italy and in a first-century votive relief from Boeotia. 98 The picture of these bottles and their contents as symbolically female attributes is further enhanced by burial evidence, although again not as an exclusive 93 For references to artistic representation, see Allison 2006a, 23;2007, 346; 2013, 100-1. For association with medical equipment, see Künzl 1983, 88-9, fig. 66; 93-4, fig. 74; Jackson 1986, 157-58;1988, esp. 74;Price 2005, 180 association. Martin-Kilcher studied more than 350 artifacts in the grave assemblages from the Late Republican and Early Imperial cemeteries in the region of Lake Maggiore, Locarno. 99 Analyses of these graves lacked sexing of the skeletal remains. However, on the basis of repeated combinations of what she argued were gender-specific artifacts in these assemblages (e.g., she considered spindlewhorls, gold and silver jewelry, brooch types, and mirrors to be female attributes, and weapons, esp. swords and lances, to be male attributes), Martin-Kilcher identified some 68 female and some 75 male burials. 100 None of the earliest burials at Ornavasso-San Bernardo (dating to the second century B.C.E.) obviously included blown-glass so-called balsamaria, or indeed ceramic so-called unguentaria, but these types of bottles (in ceramic or glass) did occur in 28 of the 48 burials at this site identified as female, and in 12 out of the 40 male burials. 101 Thus, there is a strong pattern differentiation between the two types of assemblages; the small bottles tend to be female attributes, although they are not exclusively so. Glass balsamaria are viewed as a Mediterranean element introduced into this Alpine region, Raetia, and the German provinces during the Early Empire, and predominantly in female graves. 102 These bottles are therefore also associated with changing sociocultural and grooming practices in this region during the Early Empire.
Fecher studied the graves from the Flur "Kapellenösch" cemetery at Rottweil, in southern Germany, which date between ca. 70 and 200 C.E. 103 Analyses of these graves did include sexing of skeletal remains. Twenty-five of the graves each had up to eight ceramic and glass bottles, most of which were primary grave goods. Only one bottle occurred in a grave with definite male skeletal remains: Grave 694. The other approximately 36-45 bottles were all in graves likely to have had women's burials (10 graves); in graves with juveniles (six graves); or in graves of adults of indeterminate sex (eight graves). 104 The sexing of the skeletal remains from the Rottweil graves renders the gender associations of these bottles more convincing than the gender ascriptions made in Martin-Kilcher's study. Together, though, these two studies indicate a greater propensity, at least in Early Imperial burials in southern Germany and northern Italy, for these small bottles and their contents to be more female than male attributes. Further evidence that these types of bottles were associated with female grave goods and female burials in this and other parts of the Roman world has also been noted, and Swift observed their association with the burials of wealthy women in Roman Britain. 105 Cool reported "unguent bottles" or "bath bottles" (aryballoi) from the third-century graves at Brougham; the former were used as grave goods, and the latter were associated with the cremation process. 106 This shows the widespread and continued use of these bottles and their association with burial practices, but unfortunately there is insufficient information on the sexing of the Brougham burials.
Thus, the combined literary, representational, and burial evidence, including burials with and without sexed bodies, presents a strong case that the small bottles found in archaeological contexts could be used for cosmetics and perfumes and that long, narrow unguentaria or balsamaria in particular have some female associations, at least in Italy and southern Germany during the Early Empire. While this gender attribution is by no means certain for all occurrences of these bottles across the Roman world, the combined evidence makes their characterization as predominantly female a good basis from which to interrogate gendered practices in other contexts that lack bodies but include such bottles. For example, three small, narrow bottles were found with two spindles, a bone (cosmetic?) spoon, and a small pot in Room 2 in the Casa del Fabbro in Pompeii. 107 The association of the bottles with these other artifacts implies the use of this room for women's activities of cloth working and personal hygiene, or at least for the storage of such material. This assemblage therefore potentially identifies this room as a women's space and marks it out from the more prolific evidence for industrial, and more masculine, activities at the rear of this house. 108 Another Pompeian example concerns Shop I.10.2-3, which was probably a shop for dispensing food. 109 Four small glass bottles, a larger one, and a small flask (possibly an aryballos) were found together in one corner 99 Martin-Kilcher 1998. 100 Martin-Kilcher 1998, figs. 6, 9. 101 Martin-Kilcher 1998. On the chronology of Roman blown glass, see Stern 1999. On the chronology and function of ceramic unguentaria, see Anderson-Stojanović 1987. 102 Martin-Kilcher 1998, 211; see also Fasold and Witteyer 2001, 302;Schürger 2001. 103 Fecher 2010. 104 Allison 2013, table 5.3. 105 E.g., at Wederath-Belginum, a bottle found in a possible young girl's burial apparently contained a cosmetic substance (Eckardt and Crummy 2008, 27). For this association in Roman Britain, see Swift 2011, 208. 106 Cool 2004, 365-66, 370. 107 Allison 2006a107 Allison , cat. nos. 10432008b, cat. nos. 1043 108 Allison 2006a, 342-45. 109 Allison 2006a, 297. of a room behind the shop, and further small bottles were recorded in associated disturbed deposits. 110 A graffito on the shop wall has been used to identify the owner as "Coponia," who had a maid called "Iris." 111 This is a rather unusual instance of epigraphical evidence for sexed bodies in a lived context. Laurence counted this shop among the so-called popinae in Pompeii and suggested that prostitution was likely to have taken place there. 112 As discussed above, such bottles could be used for cosmetics and medicines. It is interesting to contemplate whether their relative abundance here indicates that beauty and health care, perhaps in association with sex, were dispensed from this shop as well as food. Substantial remains of at least seven glass bottles were found within the second-century auxiliary fort at Ellingen, near Weißenberg. 113 These were among some 65 artifacts recorded inside this fort that were potentially associated with women, along with the skeletal remains of at least five, and up to 11, neonates. 114 The bottles were mainly found in the same parts of this fort as other remains associated with women, including in the soldiers' barracks. 115 They can therefore be used to support the argument for women's presence in these residences.
Thus, these types of bottles and their spatial distribution and associations with lived contexts can be used to interrogate, or to substantiate, the gender attributions of other artifacts in such assemblages. They can also be used to investigate how women's activities were integrated into domestic, commercial, and military spaces and may change our perspectives on gendered sociospatial practices in the Roman world. When these bottles occur in military contexts, particularly in the barracks of auxiliary soldiers, as at Ellingen, they are undoubtedly associated with Roman concepts of beauty, personal hygiene, and health, in contexts that were probably the domiciles of indigenous soldiers and their families. These bottles therefore provide insights into the spread of such practices among the provinces and possibly the place of local women in this adoption and spread. The examples presented here are admittedly limited but constitute useful data for investigating further comparable examples.

Artifacts Associated with Less Gender-Specific Cloth Working: Needles and Needlework
Numerous bone and metal needles have been recorded on Roman sites ( fig. 3); many were likely to have been used in cloth production and maintenance, although arguably not all. 116 For example, some of the larger standard iron needles with stout stems, found in archaeological contexts and ranging from approximately 110 to 200 mm in length, could have been used as packing needles. 117 Large standard metal needles (lgth. ca. 150 mm) could have been used in surgery, and cruder bone needles may have been used for netting or weaving (see fig. 3a). 118 Needles may also have been used for hair arranging and thus could have been part of the mundus muliebris. 119 Thus, smaller standard needles from archaeological contexts-which occur in both bone and metal and range in length from approximately 50 to 135 mm-are likely to have been for sewing but could also have been used for medical activities. 120 While there is considerable written, representational, and burial evidence that cloth working was predominantly a female task in the Roman world, this applies most particularly to spinning rather than to sewing and needlework. 121 In the written sources, the only potential reference identifying sewers as female is the use of the word "vesticiae," presumably meaning "clothes makers." 122 Certainly, in imperial households during the Early Empire, male vestifici and sarcinatores (clothes menders) were recorded. 123 Needles are difficult to depict on sculptural representation and were infrequent burial goods. 124 No needles were reported in the burial assemblages studied by Martin-Kilcher. The only Rottweil grave (Grave 162) that contained a needle (lgth. 51 mm) was likely to have been the burial of an adult female. 125 Cool recorded two iron needles in the third-century cemetery at Brougham, one from the funeral pyre of a young 110 Allison 2006a, cat. nos. 59-63, 104, 105, 121, 122. There is a lack of evidence to support an earlier suggestion that these bottles may have been used for food essences and condiments (Allison 2006a, 297, 376-77). 111 Ling 1997, 42;, 297. 112 Laurence 1994 Allison 2013, 257-58. 114 Allison 2006c2013, 263, 266-68. For the data sets, see the downloads for Ellingen in Allison 2012. 115 Allison 2013, 270. 116 For needles used in cloth working, see, e.g., Crummy 1983, 61. 117 Manning 1985, 35-7. 118 On needles for surgery, see Jackson 1994a, 204-5, nos. A14, A15; Jackson 1994b, 176-77. On needles for netting or weaving, see Crummy 1983, 65-7;Allason-Jones and Miket 1984, 61-5, cat. nos. 2.260-62. 119 Shumka 2008, 182. 120 Allison 2006aShumka 2008, 182. 121 Dixon 2007Roth 2007, 59, 89-118. For further discussion and references, see Allison 2007, 348-49;2009, 18-19;2013, 93-5. 122 See, e.g., Dixon 2000, 8. 123 Treggiari 1976 On sculptural representation, see Shumka 2008, 182. 125 Fecher 2010 person, but noted the paucity of needles in Romano-British burials. 126 Of some 50 bronze and bone needles recorded at Catterick, none would appear to have been found in burial contexts. 127 Only four iron needles were recorded in the fourth-century graves at Lankhills,in Graves 152,184,351,and 435. 128 Only one of these graves, Grave 351, had evidence of material that could potentially be gendered female, although none had male gendered items, such as crossbow brooches or any belt fittings. 129 These burial contexts hint that needles were female attributes but were not strongly gendered, at least symbolically. Contexts that lack bodies ostensibly provide more information on the actual gendered use of needles than do sexed contexts on the symbolic gendering of needles and needlework. The discovery of a bronze needle (lgth. 67 mm) within the turrets of Hadrian's Wall was used by Allason-Jones to highlight that soldiers were probably responsible for the mending and upkeep of their clothing. 130 However, as noted above, her argument was based on assumptions about the masculinity of the context. It is conceivable that women frequented these towers, perhaps illicitly, but their presence cannot be argued convincingly based on this evidence alone.
In the Lucanian villa at San Giovanni di Ruoti, four of the eight bronze and bone needles from period 3B contexts (dated ca. 460-545 C.E.) are from the area of Corridor 43, which led to the bath complex. 131 Loomweights and possibly spindlewhorls were also prominent in this area, as were items of jewelry and a bone comb. 132 This artifact assemblage implies that women congregated and worked cloth in this open and light corridor area. The needles add weight to this identification of gendered sociospatial practice but do not carry sufficient gender characterization on their own to lead to such an identification.
Of the 16 standard bone and bronze needles recorded in the Insula of the Menander in Pompeii, which range in preserved length from 50 to 118 mm, at least 11 were recorded in assemblages that appear to be of women's items. One bone needle (preserved lgth. 85 mm) was found associated with a possible spindle and gaming pieces in Room 1 of the Casa del Menandro. One bone needle (lgth. 118 mm) was recorded with jewelry and personal-hygiene equipment in a cupboard in the atrium of the Casa del Fabbro. One bone needle and six bronze needles (lgths. 50-70 mm) were recorded in Room 5 of the Casa del Fabbro along with a small glass unguentarium and what was probably a bone spindle. And one bronze needle (lgth. 56 mm) was recorded in what appears to have been a storeroom, Room 12 in House I.8.10, along with a range of material including jewelry, gaming pieces, an ear cleaner, a warp beater, and five bone implements that may have been spindles. 133 A further bronze needle associated with a glass balsamarium in Room 36 in the Casa del Menandro is noteworthy but inconclusive. 134 While the needle in Room 12 in House I.10.8 was probably in storage with other clothworking items, the one in the cupboard in the atrium of the Casa del Fabbro might be identified more securely as part of a mundus muliebris. Those in Room 126 Cool 2004, 393. 127 Cool 2002 Clarke 1979, 249. 129 Clarke 1979, 299 (bead necklace), 307 (bracelet), 316 (two hairpins).
While there is a prominent association of these needles with potential women's space in these domestic contexts in Italy, and interestingly with gaming, needles alone are not useful for identifying gendered sociospatial practices. 135 Bronze needles also occur in less female assemblages in Pompeian houses: four bronze needles (lgths. 78-93 mm) were recorded in a large assemblage in the garden portico of the Casa del Fabbro, which included measuring and woodworking equipment. 136 In the same assemblage, one bronze needle (lgth. 110 mm) was found in a concretion of what appears to be medical or toilet equipment. 137 There is no essential difference in the size and types of needles between those found in the portico of the Casa del Fabbro and those in the potential women's assemblages, with the exception of bone needles among the latter. This sample of domestic contexts is admittedly small, but it suggests a stronger female gendering of bone needles, which might confirm the main association of bone needles with sewing cloth and perhaps hair arranging. 138 Bronze needles of standard type and size also seem to have a strong association with women and cloth working, but they had a greater range of less specifically gendered uses. 139 Comparable needles have been recorded within Roman military bases. Some 29 bone and 12 bronze needles were recorded from the fort at South Shields, although without precise contexts. 140 In the fort at Ellingen, at least four bronze needles and remains of up to another six bone and eight bronze needles or pins were recorded. 141 These were found predominantly with residential buildings, Buildings B, C, and F, and tended to be associated with other female artifacts. 142 Vegetius considered that anyone involved in cloth working was an unsuitable recruit for military life, and weavers were reportedly banned from the army by law. 143 However, this prohibition may not have applied to sewing and mending. 144 In the Pompeian domestic contexts, more masculine assemblages provide evidence for the wider use of needles but not necessarily how or by whom. It is certainly noteworthy that the fort at Ellingen, which had considerable evidence for the presence of women, also had relatively large numbers of possible needles, both in bone and bronze. 145 The current evidence at these sites, and in sexed contexts, indicates that needles and their related activities had female associations but that they were less specifically gendered than the other two artifact types discussed here. Needles are, therefore, less useful for identifying gendered space and practice, but at the same time their inclusion in assemblages with other potentially female-related artifacts can support the identification of a location of female-related activities.

concluding comments
This article showcases some of the rich body of artifactual evidence from sexed and unsexed contexts that can be rigorously analyzed for a more materialcultural and gendered approach to social behavior in the Roman world. It also highlights the universality of certain types of artifacts across the Roman world that can be used more critically to investigate gendered roles in the spread of Roman cultural practices. And it demonstrates that the investigation of artifact assemblages is important for better understandings of gendered sociospatial practices.
More specifically, I argue that the consideration of different levels of gendered characterization for particular artifact types constitutes a useful interpretive tool for investigating how gender was played out in lived space in the Roman world. These three types of artifacts have uncertain gender associations. Nevertheless, they illustrate how symbolically gendered artifacts from contexts with sexed bodies can be used as a basis for more holistic investigations of actual gendered practice in lived spaces that lack such sexed bodies. [AJA 119 They also demonstrate that relationships between symbolic and actual practices need to be treated with caution. In rare instances, sexed evidence is associated with lived space, as in the graffiti in Shop I.10.2-3 in Pompeii and as can be argued from the evidence of infant burials at Ellingen. 146 Such instances provide more substantiated information on gendered sociospatial practices. In other instances, artifact assemblages in lived spaces can help us develop a better understanding of how particular artifacts may have been used and gendered and also why they might not carry symbolic gendered characteristics.
Becker argued, contra van Driel-Murray, that we need to "focus on small finds which have definite gendered associations and which exclude any exceptional usage." 147 None of these artifact types has an exclusive, or assured, gender characterization. It is doubtful that any Roman artifacts can be "definitely gendered" or that we will ever be able to exclude exceptional usage. However, more systematic and interrogative approaches to a range of different types of contexts and to analyses of particular artifact types can help identify consistent patterns of gender association. Of importance are repeated patterns of association across a number of contexts, regions, and periods. Such continuity of practice, both symbolic and actual, can serve as a basis for artifact characterization for the further exploration of gender associations, gendered practices, and gendered use of space, as well as for exploring changing gender associations and practices across different social contexts and regions. These characterizations can be used to critically examine often androcentric approaches to lived space and to how different gender and status groups interacted with material culture. 148 The inclusion of gendered perspectives in debates such as those surrounding "romanization" and "imperialism" can add critically important dimensions to our understandings of these processes. Such gendered characterizations of artifacts can also be used for more pluralist interpretative approaches to artifact assemblages. These assemblages may then be used as signifiers of gendered practices across a range of contexts and a range of types of people. 149 The aim of this study is, as stated by Roth in her volume on agricultural slavery, "not to seal particular types of evidence but to feed gender into our understanding of past societies." 150 The processes presented might be considered to represent very positivist approaches to what Allason-Jones refers to in the title of a 1995 essay as "'sexing' small finds." 151 Again, the evident patterns of habitual practice in this material and its contexts are important here, rather than how individual items might be sexed anecdotally. The associated activities and modes of dress are used to make "interpretative links between objects and social roles and identities." 152 The gender characterizations I have attempted here are sensitive to the specific contexts but not dictated by past attitudes to context function.
While these examples of artifact types illustrate some of the wealth of Roman artifactual data that can potentially be gendered, they also illustrate their often inconsistent and compromised collection and analyses. In Roman archaeology, we are often analyzing data collected in the past using quite different methods with very different research questions. The long and complex history of Roman archaeology and its disciplinary alliances has led to an unevenness in data quality as well as in engagement with feminist and gender theory. However, the above discussion demonstrates that previous studies, which have attempted to "sex" artifacts without the benefit of specifically sexed bodies, are still useful and should not be dismissed out of hand. Such studies often help substantiate recent investigations that have dealt with sex and gender more critically. Rather than throw the proverbial baby out with the bathwater, it is important to reexamine the full range of evidence as well as past interpretations of this evidence. Irrespective of exceptional cases, the resulting gender characterizations can be used for more nuanced, less androcentric, approaches to interpreting the use contexts of these artifacts, their assemblages, and the activities with which they are associated. As in the case of military bases, such an approach can be used to identify potential gender associations that "search less for certainty than for multiple plausible scenarios." 153 Much of the above argument and the gender characterizations discussed may seem self-evident and traditional. This is true, in part. However, I argue that archaeologists have recently felt constrained from using such characterizations because of the risk of seeming to stereotype gendered practice. They have often felt restricted from exploring gendered practices through the interrogation of artifacts and the archaeological record because of the lack of explicit and secure evidence for gendered behavior. However, 146 For discussion and references, see Allison 2013, 261-65. 147 Becker 2006, 37;van Driel-Murray 1997, 55. 148 See, e.g., Gardner 2007, 82. 149 Cf. Gardner 2007, 82, 89. 150 Roth 2007, 58. 151 Allason-Jones 1995 Stig Sørensen 2006, 28-9. 153 Tomášková 2006, 25. feminist material-cultural approaches, especially to the interpretation of artifacts, combined with critical, interrogative analyses of textual, epigraphical, figurative, and burial evidence where sexed bodies are present, can be employed to explore gendered associations. While many gender characterizations of artifacts may be shown to be incorrect, through further study they can form a useful first step for developing genderbased analyses in Roman archaeology, especially in contexts were no actual sexed bodies are present.
The examples discussed here have been investigated through published material, often published within quite different and traditional scholarly frameworks. Allason-Jones stressed the need for careful and critical approaches to such artifact catalogues, which have not always published all artifactual evidence and its contexts. 154 She reiterated Bishop's emphasis on the need to "feel comfortable about the taphonomy" 155 and stressed the desirability of the firsthand study of artifacts and artifact assemblages. Allason-Jones' comments and the analyses in this article highlight the need for more rigorous, more holistic, and more contextualized cataloguing of artifacts and also for more detailed publication that considers artifact consumption as well as production. 156 Especially appropriate for the publication of excavated artifacts are online, open access data resources that give scholars much greater access to such material 157 and also reduce the need for selectivity due to the expense of paper publication. Better taphonomic information in more recent excavations is certainly helping develop more contextualized approaches to artifacts and artifact distribution. Unfortunately, there are many Romanperiod sites from which the available information is less precise and less fully presented but from which better-preserved artifacts and artifact assemblages have often been excavated. Particularly relevant here are artifacts from rapidly abandoned military sites, and of course Pompeii, where the types of artifacts discussed above could often have been lost or abandoned in their place of use. Again, such contexts and such material do not provide ideal data, but the extensiveness of such material and its availability for study mean that quantitative comparisons for consistent patterns of practice can often compensate, at least in part, for taphonomic uncertainties.
Finally, this article aims to present approaches, analytical tools, and some case studies that can help increase "conversations between social and material traces of the past." 158 It also aims to exemplify how more integrated approaches to Roman artifacts and their gender associations can make a greater contribution to the fast-moving field of gender studies related to the ancient world. At the same time, it highlights the complexity of these data and of their investigation. Stig Sørensen asked whether Roman archaeology can contribute to how we investigate gender more broadly or whether it merely uses principles from other branches of archaeology and the social sciences. 159 The interrogative processes outlined in this article, based on a wealth of data with good historical specificity, can contribute to greater understandings of the histories of various gendered practices that are relatively free of undue stereotyping.