Vuolanto ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 2017 Ville Vuolanto Grandmothers in Roman Egypt Ville Vuolanto CHOLARS OF THE ROMAN FAMILY have long acknowl- edged grandmothers’ role as “useful supplementary carers or custodians” in the absence of parents.1 How- ever, there is hardly any discussion of the roles of grandparents in the everyday life of children. Only in some studies, dedicated to old age, women, or children, do we find more than a passing remark on the significance and possible agency of the grand- parents—and even here, the discussion most often concerns the elites, or the imperial family.2 1 B. Rawson, Children and Childhood in Roman Italy (Oxford 2003) 239–241 (quote), with esp. S. Dixon, The Roman Mother (London 1988) 33, 76–84, 132–133, 217–220; T. Parkin, Old Age in the Roman World: A Cultural and Social History (Baltimore 2003) 49–54; and K. Cokayne, Experiencing Old Age in Ancient Rome (London 2003) 166–170. 2 C. Laes, “Grandmothers in Roman Antiquity: A Note on Avia Nutrix (AE 2007, 298),” Melita Classica 2 (2015) 99–113; V. Vuolanto, Children and Asceticism in Late Antiquity. Continuity, Family Dynamics and the Rise of Christianity (Ashgate 2015) 172–174, 186–188, 199; A. Pudsey, “Children in Roman Egypt,” in J. Evans Grubbs et al. (eds.), The Oxford Handbook of Childhood and Education in the Classical World (Oxford 2013) 494–500; C.-E. Centlivres- Challet, “Pliny the Nephew: Youth and Family Ties across Generations and Genders,” in M. Harlow and L. Larsson Lovén (eds.), Families in the Roman and Late Antique World (London 2012) 9–13; T. Parkin, “The Roman Life Course and the Family,” in B. Rawson (ed.), A Companion to Families in the Greek and Roman Worlds (Malden 2011) 286–289; C. Baroin, “Ancestors as Models: Memory and the Construction of Gentilician Identity,” in V. Dasen and T. Späth (eds.), Children, Memory, and Family Identity in Roman Culture (Cambridge 2010); P. Gallivan and P. Wilkins, “Familial Structures in Roman Italy: A Regional Approach,” in B. Rawson and P. Weaver (eds.), The Roman Family in Italy: Status, Sentiment, Space (Oxford 1997) 250–252, 267– S VILLE VUOLANTO 373 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 In research on past societies that is influenced by anthropol- ogy, demographic theory, and evolutionary biology, however, the role of grandmothers has occupied a central position. In particular, the ‘grandmother hypothesis’ has aroused interest: it is proposed that rather than continuing to have their own chil- dren, older women, especially maternal grandmothers, would contribute more to the survival and transfer of their line (and thus of their genes) by taking care of their younger female relatives and their children. This would have contributed to the development of a longer post-menopausal life span for women, and would have led humans to elaborate distinctive family structures and behaviour, especially the cooperative caring for offspring. Indeed, many historical and anthropological studies have shown a positive link between child survival and the presence of (maternal) grandmothers.3 Although the material from the Roman world cannot supply this discussion with quantitative data about survival rates, it is possible to study the agency and compare the relative significance of different grandparents in the early lives of children. This would help us assess the plausibility of the some specific claims made by the hypothesis, especially the proposal that maternal grandmothers would appear more often in caring roles than paternal grand- mothers or grandfathers. I concentrate here on grandmothers and their interaction ___ 268; J.-U. Krause, Witwen und Waisen im Römischen Reich III Rechtliche und soziale Stellung von Waisen (Stuttgart 1995) 57–62, 67–73. 3 See esp. R. Sear, N. Allal, and R. Mace, “Family Matters. Kin, Demog- raphy and Child Health in a Rural Gambia Population,” in G. Bentley and R. Mace (eds.), Substitute Parents. Biological and Social Perspectives on Alloparenting in Human Societies (New York 2009) 50–76, esp. 51–52 and 67–68; M. Gur- ven and H. Kaplan, “Longevity Among Hunter-gatherers: A Cross-cultural Examination,” Population and Development Review 33 (2007) 350–352; E. Voland, A. Chasiotis, and W. Schiefenhövel, “Grandmotherhood: An Overview of Three Related Fields of Research on the Evolutionary Signifi- cance of Postgenerative Female Life,” in E. Voland et al. (eds.), Grand- motherhood: The Evolutionary Significance of the Second Half of Female Life (New Brunswick 2005), with the other contributions in the same volume. 374 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 with grandchildren. However, these activities are to be con- textualized with the help of information on grandfathers and with other adult relatives of underage children. In what kind of situations were grandparents in charge of their grandchildren? Did it require the absence, or even the death, of the parents? In what ways did grandmothers influence the lives of their family groups, and what kind of authority did they have? In the everyday life of minors, especially of orphans, was it merely the availability of certain relatives that mattered—or were there certain tasks or roles especially reserved or promoted for (maternal or paternal) grandmothers in preference to other relatives? To shed light on the grandmothers’ roles in the inner dy- namics of households and in the lives of their grandchildren, I use three sets of sources from Roman Egypt. Most centrally, I use papyrological documentary sources pertaining to the Hel- lenised metropolis Oxyrhynchos. A systematic reading of these sources yields 34 cases which feature grandparents with under- age children, dating from the first century CE to the mid-sixth century.4 I have included cases in which children who either are said explicitly to be fourteen or younger, or who can reasonably be supposed to have been minors, appear with their grandparents. This sample of grandparents covers a negligible proportion of the overall body of circa 5600 edited documents from the whole Oxyrhynchite nome, and still a very small pro- portion of the total of circa 500 documentary papyri which include a reference to children and childhood. However, since the sample comes from a systematic reading of all the Oxy- rhynchos cases, this will give the best possible basis for com- 4 All Oxyrhynchos-related papyri are included (both the town and the nome), indicated as such in www.papyri.info by the end of 2015. Twenty- one of the cases involve grandfathers; 9 involve grandmothers; in 3 cases there appear both a grandmother and a grandfather, and in 1 case it is un- clear whether the grandparent in question is male or female. There are also 2 step-grandparents. VILLE VUOLANTO 375 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 paring the activities of grandfathers and grandmothers in one limited geographical area. The second set of materials consists of the census documents for taxation from Roman Egypt, recording living arrange- ments. Those concerning grandparents appearing with their grandchildren have been collected by April Pudsey, and they include 19 grandmothers, 1 great-grandmother, and 2 step- grandmothers.5 The third group of sources consists of those cases in con- temporary papyri outside Oxyrhynchos which mention grand- mothers. This third set, however, is not based on a systematic reading of the edited papyri, unlike the two other sets of ma- terial, nor does it claim to be comprehensive: these 22 cases are those I have come across in earlier studies and in my own research. The paucity of cases reflects the basic demographic patterns of the Roman world; it is clear that only a minority of children had any chance of knowing their grandparents. Theoretically, a Roman child had approximately a four-in-five chance of having any of his grandparents alive at birth, but this would have dropped to only a one-in-two chance by the age of ten.6 On the other hand, according to the calculations of Sabine Huebner, some 15% of recorded children lived in three-gen- erational households in Roman Egypt—and, in this material, there appears no evidence of children living together with their grandparents without any relatives of the intermediate genera- tion.7 Nevertheless, children in many households were in direct 5 Pudsey, in The Oxford Handbook 491, 497–498. See also Table 1 below. 6 This follows from the tables generated with the help of the model life tables in R. Saller, Patriarchy, Property and Death in the Roman Family (Cam- bridge 1994) 49 and 52. On the problems of model life tables see W. Scheidel, Death on the Nile: Disease and the Demography of Roman Egypt (Leiden 2001) 118–142; they nevertheless provide the necessary rough estimates of the scales and relative effects of different demographic circumstances. 7 S. Huebner, The Family in Roman Egypt (Cambridge 2013) 73. 376 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 contact with their grandparents,8 and it has to be remembered that the 15% means at a certain point of time—more (younger) children would have been living with grandparents at some earlier point of the family life course. 1. Family structure and cohabitation To give an overview of the grandmothers’ roles in families, I begin with a study of the patterns of cohabitation. How often did children live with their grandparents? Are there any pat- terns to be discerned when comparing paternal and maternal grandfathers and grandmothers with regard to living arrange- ments? The virilocal marriage pattern in Egypt meant that newly- married women tended to move into their husbands’ family households.9 In everyday contexts, therefore, there is a higher probability that small children were in contact with their paternal than with their maternal relatives. On the other hand, there was a pattern of younger women marrying older men— Roger Bagnall and Bruce Frier have calculated that for women aged thirteen to nineteen years, the age gap between them and their spouses is nearly nine years.10 It was more probable to become a maternal grandmother at the age of 45 than a paternal grandfather at the age of 55. Most importantly, at the age of 10, a child would have had a one-in-five chance of having a living grandfather, but a two-in-five chance of having a living grandmother. The probability of having a maternal grandmother living was twice that of having a paternal grand- mother or maternal grandfather, and seven or eight times higher than the probability of having a living paternal grand- 8 Parkin, Old Age 52. 9 Huebner, The Family in Roman Egypt 48–50. 10 R. Bagnall and B. Frier, The Demography of Roman Egypt 2 (Cambridge 2006), esp. 118–121. The average age gap for all married couples would have been 7.8 years. Although the reliability of the exact figure may be in question, the trend for women marrying older men was prominent: see Huebner, The Family in Roman Egypt 31–49, 58–59, 92–106. VILLE VUOLANTO 377 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 father. All this would have shifted the balance in the direction of maternal relatives, especially later in childhood: paternal grandparents, especially grandfathers, would more probably have been dead when grandchildren needed help if, for exam- ple, their parents were ill or deceased. Theoretically, therefore, if the relationships between children and their grandparents had been determined only by demography, we should see in our sources a clearly higher number of active grandmothers than grandfathers in contact with their grandchildren.11 To a certain degree, the actual ages of grandparents found in the census records fit the theoretical model. Women, who mar- ried earlier, could have become grandparents quite young. The youngest grandmother who appears in the census documents is only 42 years old, and the youngest grandfather 54, while the oldest grandparents of both sexes are 75 years old (this co- incidence is probably due to age rounding). The average age is 60 for grandmothers and 63 for grandfathers—not any great difference here.12 Although these numbers by no means nullify Mary Harlow’s conclusion that becoming a grandparent did not necessarily mean growing old in years, but rather marked a growing maturity within the family,13 the papyri show that even in the ancient demographic regime, there certainly were grandparents of an advanced age who were still in contact with their grandchildren—and thus, from the viewpoint of the chil- dren, they did indeed represent old age and past generations. 11 According to the tables in Saller, Patriarchy, Property and Death 49 and 52 (see n.6 above on model life tables). On marriage and demography in Roman Egypt see A. Pudsey, “Nuptiality and the Demographic Life Cycle of the Family in Roman Egypt,” in C. Holleran and A. Pudsey (eds.), Demog- raphy and the Greco-Roman World. New Insights and Approaches (Cambridge 2011) 60–98. 12 However, there are only fifteen grandmothers and ten grandfathers in total (material from Pudsey, in The Oxford Handbook 497–498). 13 M. Harlow, “Blurred Visions: Male Perceptions of the Female Life Course: The Case of Aemilia Pudentilla,” in M. Harlow and R. Laurence (eds.), Age and Ageing in the Roman Empire (Portsmouth 2007) 208. 378 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 Lineage Sex Number of grandparents by different data sets Oxyrhynchos Census records Other papyri Paternal grandmother 6 11 12 grandfather 8 12 n/a Maternal grandmother 5 6 9 grandfather 14 2 n/a Both sides grandmother 0 2 0 grandfather 1 2 n/a Unclear grandmother 1 0 1 grandfather 0 0 n/a unclear 1 0 n/a TABLE 1: Maternal and paternal grandparents In documents from Oxyrhynchos, we find considerably more maternal grandfathers (14) than any other types of grand- parents: 5 maternal grandmothers, 6 paternal grandmothers, and 8 paternal grandfathers (see Table 1).14 In addition, in one text it is unclear whether the grandmother is paternal or maternal, and in another there is a paternal grandparent of unknown sex. In one further case, the grandfather is simultane- ously maternal and paternal (because of a sibling marriage).15 14 Paternal grandmothers: P.Oxy. III 479; P.Fouad 35; P.Oxy. LIV 3754; P.Oxy. LVI 3860; P.Oxy. III 496; PSI III 236. Paternal grandfathers: P.Oxy. XXXVI 2787; SB X 10220; P.Oxy. XXXIII 2671; PSI I 53 col. i; P.Oxy. LXXVIII 5182; P.Giss. I 34; PSI III 236; P.Oxy. I 54 and VI 908 (these two cases concern the same individuals). Maternal grandmothers: P.Oxy.Hels. 29; P.Oxy. LIV 3770; P.Coll.Youtie 67; P.Oxy. XXXI 2601; P.Oxy. XLIX 3491. Maternal grandfathers: SB VI 9421; PSI XII 1247; P.Oxy. LXVII 4585; P.Coll.Youtie 67; P.Oxy. XII 1581; P.Oxy. III 491; P.Oxy. XXXIV 2709; P.Oxy. VI 906; P.Oxy. XIX 2235; P.Oxy. LIX 3998; P.Mich. XVIII 789; P.Oxy. XLIX 3491; P.Oxy. I 68; P.Oxy. I 105. 15 Paternal or maternal: P.Oxy. XIV 1679; unknown sex: PSI III 210. Sib- ling marriage: P.Oxy. XII 1452. VILLE VUOLANTO 379 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 Taken together, there appear 12 grandmothers and 23 grand- fathers—thus, the impression we have from earlier studies and from demography, that we should meet more grandmothers than grandfathers in our sources, is challenged.16 When we shift the focus to the more random set of grand- mothers outside Oxyrhynchos, the situation does not change much, although we find slightly more cases with paternal than with maternal grandmothers (see Table 1).17 However, the census records with grandparents again record a different situa- tion: while in total numbers there are no surprises, since there appear 19 grandmothers and 16 grandfathers, the relative oc- currence of paternal and maternal grandparents varies greatly: 23 paternal grandparents but only 8 maternal grandparents (Table 1). Even the one great-grandmother who appears in my sources is paternal.18 Thus, only one-fourth of the grandparents involve maternal grandparents.19 The difference is at its most extreme with the grandfathers: there appear only 2 maternal, but 12 paternal grandfathers. The huge difference between the visibility of paternal and maternal grandparents might at first sight invite explanations based on local circumstances, but it can be explained by the different focus of the separate sets of sources. The census material reflects, by definition, co-habitation patterns, whereas 16 Parkin, Old Age 51–56; Pudsey, in The Oxford Handbook 494. 17 Paternal grandmothers: PSI X 1159; SB V 7558; SB X 10571; SB XX 15188; P.Lips. I 28; SB V 7737; P.Kell.Copt. 11; P.Mil.Vogl. I 25; P.Brem. 63; BGU II 380; SB XVI 12981; BGU III 948; Maternal grandmothers: P.Vindob.Tandem 27; P.Diog. 11/12; P.Mert. II 63; P.Mich. VIII 514; P.Lond. III 951 v. (less likely, the writer could also be a grandfather); P.Brem. 63; SB XVI 12981; BGU III 948; P.Col. VIII 215. In SB V 7572 the grandmother may be paternal or maternal. 18 In record 201-Ar-9 (references to census households follow the cata- logue by Bagnall and Frier, Demography). 19 In two further cases the grandparents are simultaneously maternal and paternal (because of a sibling marriage): 187-Ar-4 and 159-Ar-11 (here even the grandparents were siblings). See also Parkin, Old Age 53. 380 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 the other materials would document patterns of agency and interaction. Indeed, even among the cases from Oxyrhynchos, the three census documents all include paternal grandfathers (and one paternal step-grandmother). Of the other cases from Oxyrhynchos which show cohabitation of grandchildren with their grandparents, only one refers to a paternal grandfather who seems to live with his grandchildren, while in two cases there appear paternal grandmothers living with underage chil- dren.20 There is a maternal grandmother who permanently lives with her grandson, and a boy who seems to have lived with his maternal grandfather. In two other cases, grand- children were staying at the homes of their maternal grand- fathers, but these seem to reflect temporary arrangements.21 In the cases involving grandmothers outside Oxyrhynchos, the prevalence of cohabitation with paternal relatives emerges con- spicuously: in six out of seven cases where it seems reasonable to assume that grandchildren were cohabiting with their grand- mothers, the grandmother is paternal.22 In all, not many children may actually have lived with their grandparents, but when they did, living with paternal grand- parents was clearly more common than with maternal grand- parents. On the other hand, it has to be pointed out that there actually were grandparents, also grandmothers, living together with and taking care of their grandchildren and their education all by themselves, both in more temporary situations when the parents of the child were still alive, and, more permanently, when they had died.23 20 P.Oxy. III 479; P.Fouad 35; P.Oxy. XXXVI 2787. 21 P.Oxy.Hels. 29; P.Oxy. LXVII 4585; SB VI 9421; P.Oxy. XII 1247. 22 BGU III 948; BGU II 380; P.Brem. 63; P.Mil.Vogl. I 25; P.Kell.Copt. 11; P.Lips. I 28. A last case is ambiguous, and it may be that we have a paternal grandmother even here: SB V 7572. 23 Cf. Laes, Melita Classica 2 (2015) 106–108. See P.Fouad 35 and P.Oxy. Hels. 29, probably also with, e.g., SB XX 15188 and PSI X 1159. In P.Brem. 63 (and P.Giss. 21, 23, 24, 77, and 78; P.Alex.Giss. 58) a grandmother takes care of her granddaughter and her studies while her parents are away. VILLE VUOLANTO 381 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 2. Grandmothers: the legal point of view Grandfathers appear in many Egyptian documents with their grandchildren in the role of guardians—a common phenom- enon across the Roman Empire.24 Traditionally, mothers were excluded in Roman law from any kind of authority position: they were not allowed to represent others in court or in finan- cial transactions. This made it impossible for women to take responsibility for any sort of guardianship. Most importantly, this was also true with regard to the guardianship of their own children and grandchildren: up to the age of fourteen for boys and twelve for girls, they could not administer their own busi- ness, and thus, if their father had died, they needed a guardian to act for them in economic and legal matters, and to take care of their everyday needs. With the age of puberty, a person sui iuris attained his or her full legal capacity. Nevertheless, it was often held that children from their early teens to twenty-five years should not achieve full economic independence, but should be guarded by a curator. Guardianship was the task of nearest adult male relatives of the child.25 However, mothers and grandmothers did take care of the affairs of their children and grandchildren. This was both be- cause of practical arrangements in families and households, and because the local understanding of the proper role of women differed from the traditional Roman view. In first- 24 Cases from Oxyrhynchos: P.Oxy. III 491; P. Mich. XVIII 789. In P.Fouad 35 there appears a step-grandfather as a guardian. See also e.g. P.Oxy. XIX 2235. For cases outside Oxyrhynchos see Krause, Witwen und Waisen III 58–61. 25 V. Vuolanto, “Women and the Property of Fatherless Children in the Roman Empire,” in P. Setälä et al., Women, Wealth and Power in Roman Empire (Rome 2002) 211–218; J. Gardner, Family and Familia in Roman Law and Life (Oxford 1998) 244–249. On cura see Saller, Patriarchy, Property and Death 188– 189, and J. Beaucamp, Le statut de la femme à Byzance (4e–7e siecle) I (Paris 1990) 101 and 319. On guardianship in Roman Egypt see C. Kotsifou, “Papyrological Perspectives on Orphans in the World of Late Ancient Christianity,” in C. Horn and R. Phenix (eds.), Children in Late Ancient Christianity (Tübingen 2009) 353–362. 382 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 century CE Egypt, for example, a grandmother was nominated in her husband’s will as guardian (epitropos).26 Later, grand- mothers still feature prominently in supervisory roles towards the children and their possible official male guardians. There is a case, for example, in which a grandmother from Ptolemais Euergetis “on her own responsibility” (τῷ ἰδίῳ κινδύνῳ) asks a strategos to nominate guardians for her three (paternal) grand- sons, two of whom are still minors—she would herself serve un- der a title of epakolouthetria to supervise the administration of the property left to the children by their deceased father, her son. She also takes the opportunity to ask if one of the minor chil- dren could be appointed to the office of kosmeteia. It seems that she had already acted as a supervisor for the children, and now some urgent need—a court case?—had made the presence of official male guardians inevitable. In another case a father had in his will expressed the wish that two men would act as guar- dians (epitropoi) for his little daughter, while the child’s grand- mother, Valeria Sempronilla, would act as an epakolouthetria.27 In one case, a woman had acted as a phrontistria (supervisor or guardian) for her grandson: it seems that his father had died a year earlier, and the grandmother had made a deal with a nurse—but now, it was necessary to revise the deal and take a further loan; for all this, the child would be in need of an official guardian. Another grandmother, Taharpagates, acted as a phrontistria for her minor grandson Pacusis, and bought a house with a garden from the boy’s mother and gave it to the boy. In this case, both the mother of the son and her new husband were alive, and presumably the boy lived with them. It seems that the grandmother wanted to make sure that the boy would have his share of the maternal inheritance even when the mother had remarried. It was, indeed, specifically pointed out that the boy’s mother had inherited the house in question from her own mother. This is also one of the two 26 P.Vindob.Tandem 27 (technically a donatio mortis causa). 27 PSI X 1159; SB V 7558. VILLE VUOLANTO 383 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 cases in which the grandmother acted as the de facto guardian of her grandchild, while the husband of the grandmother is men- tioned—in the other case, the guardianship of the property is in fact transferred to the man. Clearly, these husbands were not the grandfathers of the children in question, but new husbands of the remarried grandmothers.28 In one case, a grandmother is nominated in the will of her daughter to guard (phrontizein) her minor son with her son-in- law, the boy’s father: they are together to provide support for his nurse.29 This is a curious case: since the father was alive, no other person should be able to intervene to child’s affairs. Out- side of the juridical sphere, however, the case makes sense: the mother aims at strengthening the ties of her son also on the maternal side, ensuring the protection (and mutual supervision) of the two lineages of the boy. In four cases from Oxyrhynchos, grandmothers take action that implies authority over the grandchild, although without any particular designation: in one of these cases, a grand- mother gives a girl in marriage together with the girl’s father, her son; in another, a grandmother registers a boy in the absence of the his father, her son; in a third, a grandmother registers a boy’s move into her house.30 The common element in these three is that the fathers of the children in question are still alive. In the fourth case, it is possible to see a preference in the selection of a person to take responsibility for an orphaned child: Apollonous gives her daughter’s son as an apprentice to Apollonous’ other son, while still retaining for herself the duties of feeding and clothing the boy. The maternal grandmother was preferred as the person to take charge of the boy, not his maternal uncles.31 In yet another case, a grandmother was in 28 SB XX 15188; SB X 10571; P. Fouad 35. 29 P.Diog. 11/12 (technically a donatio mortis causa). 30 P.Oxy. III 496; P.Oxy. LIV 3754; P.Oxy. III 479. 31 P.Oxy.Hels. 29; another uncle is guardian (kurios) of his mother. See also P.Vindob.Tandem 27, with a grandmother as guardian instead of her brother, a grand-uncle of the children. 384 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 charge of her grandson, but gave him in adoption to her son, the boy’s paternal uncle. According to the agreement, both the paternal and maternal inheritances would follow the boy, and he will inherit from his uncle.32 In all these instances, the grandmother supersedes her own son or brother as a guardian. However, we do not know any cases in which a maternal grandmother would have been preferred to a paternal male relative. Even more generally, appointing a maternal grand- parent to be a guardian was by no means self-evident: in one case, the maternal grandparent does not act as a guardian, but instead a man whose relationship with the child in question is unclear. In another case, a widow asks a strategos to select guardians for her children—while her father, a maternal grandfather to the children, acts as her guardian.33 Thus, in official contexts, the maternal relatives seem to have been disfavored as the persons to take action on behalf of underage children. It also seems clear that step-(grand)parents were secondary when guardians and protection for underage children were sought. In one instance a grandmother had acted as a super- visor (phrontistria) in preference to a stepfather and a step-grand- father. In another case, Thaesis, who guarded (epitropeuein) her granddaughter Tathoonas’ possessions, gave her new husband the right to collect income and sell the girl’s property—and to appear for Thaesis in court. Even if part of the administration of the girl’s goods was passed to male hands, Thaesis was the primary guardian.34 On the whole, grandmothers seem to have been preferred for taking care of the children and their businesses rather than their (even male) relatives, but only inside of their own lineage. 32 P.Lips. I 28. 33 P.Coll.Youtie 67; P.Oxy. XXXIV 2709. 34 SB X 10571; P.Fouad 35. On stepfathers see S. Huebner, “Callirhoe’s Dilemma: Remarriage and Stepfathers in the Greco-Roman East,” in S. Huebner and D. Ratzan (eds.), Growing Up Fatherless in Antiquity (Cambridge 2009). VILLE VUOLANTO 385 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 Grandmothers were active when it was necessary to guard the interests of her side of the family—but maternal grandparents were not to take responsibility in preference to paternal (male) relatives. Moreover, in cases in which the fathers, or both parents, were temporarily away, the grandmothers were the obvious and most natural resource for the care of underage children. Many grandmothers, if their children had died, as- sumed the position of a supervisor towards their grandchildren —thus representing their line and kin group in the lives of their grandchildren. Unlike earlier scholars, I would be reluctant to call these phrontistria, epakolouthetria, and other supervisors of minors and their property ‘unofficial guardians’, or to label the phenomenon a ‘quasi-guardianship’, since the practice clearly was outside of any Roman system of legal guardianship.35 Often grandmothers took care of the children and their businesses without any formal status. Many cases come to our attention exactly because there was a change in circumstances taking place and an official (male) guardian was needed. If no such in- tervention was to take place, there was no documentation; this would mean that the cases involving grandmothers docu- mented in papyri allow us to see only the tip of the iceberg of what their actual influence seems to have been. 3. Life course and grandparents It would be natural to expect especially the youngest children to interact with their grandparents—the older the children were, the more probable it was that their grandparents had died. Can we see this pattern in our sources? Moreover, do we have here a difference between paternal and maternal grand- parents, and does the pattern of relationships change when 35 On epakolouthetria see now L. Gagliardi, “La madre tutrice e la madre ἐπακολουθήτρια: osservazioni sul rapporto tra diritto romano e diritti delle province orientali,” Index 40 (2012) 423–446. Vuolanto, in Women, Wealth and Power 203–243, Krause, Witwen und Waisen III, esp. 71–72, and Beau- camp, Le statut de la femme 314–319 and 325–337, use the terminology of unofficial/quasi-guardianship. 386 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 children get older? The cases in census records do in fact document quite young children with their grandparents: of the 55 children who appear with their grandparents in families with underage children, the average age is 5.8, and the median age only 4.0.36 What is noteworthy here is that for the paternal grandparents, the numbers are even lower: average age 4.3 and median 3.5, and only one child37 with a paternal grandfather is older than 6 (he is 14). On the other hand, of children with maternal grand- parents (together 12), the average age is 11.1 and the median age 11.5. Clearly, maternal grandparents tend to be recorded with older children than do paternal grandparents.38 This difference is to be expected in a situation where a virilocal mar- riage pattern dominates, since it was much more probable that grandchildren would be born in the households of their pater- nal than their maternal grandparents. However, the influence of maternal grandparents would last longer because of the demographic circumstances. Attestation in the census records does not depend on the age of the children—it was enough that the children and grand- parents existed, and they were documented, regardless of their possible agency or roles in the family group. As for the other documents, however, the more active the protagonists would have been, the more frequently they would have appeared in the papyri. Thus, we would expect that the age distribution of children in other kinds of documents would feature older children more prominently than younger: the older ones were 36 Based on of the list in Pudsey, in The Oxford Handbook 497–498 (with, however, grandparents in 187-Ar-4 classified as both maternal and pater- nal, since the parents of the children are siblings). 37 131-He-4. 38 In those cases (187-Ar-4 and 157-Ar-11) in which the parents of the children are siblings and thus the grandparents are simultaneously both maternal and paternal, the ages of the children mentioned are 1, 1, 13, 1, and 2—following the pattern found in census documents with paternal grandparents. VILLE VUOLANTO 387 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 more capable of individual action and thus more likely to be mentioned in documents like court actions, parental wills, epi- krisis and marriage documents, and contracts of different sorts. Unfortunately, in cases outside of the census records direct information about the children’s ages is very seldom available; in fact, we have explicit information about this in only three cases, with children aged 8, 10, and 13.39 However, one can categorize cases into three rough age groups: infants (less than about 3 years of age: six cases), children older than this but not more than about 8 (three cases, at least six children); and those older than about 8 (seven cases).40 One may note in these cases a nearly equal distribution of paternal and maternal grand- parents for the children past infancy, a feature that char- acterizes these data also more generally—but for infants, the number of interventions on the part of maternal grandmothers is conspicuous, since some other type of grandparent (a pater- nal grandmother) appears in only one of these cases. Moreover, even in two of the three ‘middle category’ cases, there appear 39 P.Oxy. III 479; P.Lips. I 28; P.Oxy. LXVII 4585. 40 Probable infants: P.Oxy. LIV 3770 (maternal grandmother); P.Lond. III 951 v. (maternal grandmother [or a maternal grandfather]); SB XX 15188 and P.Diog. 11/12, a child with a nurse (paternal and maternal grand- mothers); P.Col. VIII 215 (maternal grandmother visiting a girl three times a day; probably an infant with a wet-nurse); P.Brem. 63 (a grandmother with two grandchildren: she is the maternal grandmother to an infant, and the paternal grandmother of the other, who is “studying,” and thus older than ca. 8). The second age category: P.Oxy. XLIX 3491 (maternal grandmother: after nine years of marriage the oldest child is 8 at most); P.Oxy. VI 906 (maternal grandfather, three children, the oldest less than 10, since the mother is 24); P.Oxy. III 479 (paternal grandmother, child 8 years). The third category (with the girl from P.Brem. 63 above): P.Lips. I 28 (paternal grandmother, child 10 years); P.Oxy.Hels. 29 (maternal grandmother; child an apprentice so older than 8); P.Oxy. LXXVIII 5182 (paternal grand- children, one of whom can write a letter); P.Oxy. LXVII 4585 (maternal grandfather, child 13 years); P.Oxy. XII 1452 (grandfather, simultaneously paternal and maternal, having died “some time ago”; this is an epikrisis declaration of the boy); P.Oxy. III 496 (paternal grandmother organising a marriage for a girl). 388 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 maternal grandparents, and the children in these cases may well include infants. What we see here seems to have been caused by the com- bination of demographic and cultural factors. Since women married earlier than men, and the newly-married couple tended to move into the household of the husband, in their infancy grandchildren were most probably living with their paternal relatives—but on average, they would have more grandparents living from their maternal side. Thus, in census documents paternal grandparents would be more often men- tioned with infants and small children. However, because the paternal grandparents would have died earlier in a child’s life course, and because, at the same time, these virilocal extended families would have been dissolving, the maternal grandparents would have reached an equally important place in the lives of their grandchildren. Regardless of the cohabitation patterns, daughters would seek support from their own mothers when about to give birth and when they had small children; accord- ingly, even if maternal grandmothers appear only rarely in census material, they dominate the interaction with infant grandchildren.41 The agency of grandmothers towards their younger relatives is not in the least limited to situations in which the mother and/or the father of the child was dead.42 4. Boys and girls A further question to be considered is the potential effect of the gender roles: do there emerge any gender-related patterns with regard to girls and boys mentioned together with their 41 See e.g. SB XII 12606 (Oxyrhynchos) and P.Fouad 75 (Arsinoite) on a brother and a sister informing their mother about her daughter’s pregnancy and childbirth: in the first case the parturition has started all too early, in the second the daughter had died while giving birth to a dead child. 42 Cf. Krause, Witwen und Waisen III 67–73, and Laes, Melita Classica 2 (2015) 104. See e.g. P.Oxy. III 479; P.Oxy. LIV 3770; P.Coll.Youtie 67; P.Oxy. XXXIV 2601; P.Oxy. LIV 3754; P.Oxy. III 496 (all these papyri from Oxy- rhynchos); P.Vindob.Tandem 27; SB X 10571; P.Diog. 11/12; P.Kell.Copt. 11; P.Mil.Vogl. I 25; BGU II 280; P.Col. VIII 215. VILLE VUOLANTO 389 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 grandparents? Again, we must begin by analysing the census returns. Demographically, one would expect a roughly even distribution of boys and girls, and this is indeed what we encounter: 27 boys and 24 girls appear with their grandparents (and one girl with her great-grandparent); of these, 18 boys are mentioned with their grandmothers and 13 with grandfathers; likewise, 15 girls are mentioned with grandfathers and 14 with grandmothers. There are no ‘missing females’ here.43 However, a different pattern emerges when we turn to the documents other than the census records. In those cases in which the sex of the child in question is known to us, we en- counter 23 children, 17 boys and 6 girls. Of the girls, 2 are mentioned with grandmothers, 4 with grandfathers; 6 boys with grandmothers, 12 with grandfathers. There is no sign of gendered coupling, such as girls appearing more often than boys with their grandmothers. However, the lack of girls is striking—why they do not appear with their grandparents as often as boys? This seems to have nothing to do with the local circumstances in Oxyrhynchos, since the same pattern emerges in the sample of papyri with grandmothers outside of Oxyrhyn- chos: we find there 16 boys and 4 girls. The same phenomenon is also found when we study minors appearing with their uncles and aunts, with almost three times more nephews than nieces in cases from Oxyrhynchos.44 When we examine the cases where girls do appear, it is hard to detect any pattern. In Oxyrhynchos, we have a paternal grandmother giving de facto guardianship of a minor girl’s property to her new husband; a maternal grandfather worrying over his granddaughter’s financial interests; another maternal grandfather taking care of his two orphaned granddaughters; a 43 See R. Bagnall, “Missing Females in Egypt,” SCI 16 (1997) 121–138, with Scheidel, Death on the Nile 149–150 and 167–169. 44 A. Pudsey and V. Vuolanto, “Being a Niece or Nephew in an Ancient City. Children’s Social Environment in Roman Oxyrhynchos,” in C. Laes and V. Vuolanto (eds.), Children and Everyday Life in the Roman and Late Antique World (London 2016) 80–96. 390 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 paternal grandfather helping his granddaughter in a lawsuit; and a paternal grandmother giving a granddaughter in mar- riage (together with the girl’s father).45 Of the grandmothers outside of Oxyrhynchos, we meet Valeria Sempronilla, super- visor of her underage granddaughter; a woman writing to her son reminding him that her granddaughter needs to use shoes because of scorpions; Eudaimonis, who is taking care of her granddaughter and her studies while her parents are elsewhere; and Thermouthis, who is taking care that her infant grand- daughter is adequately nursed while the girl’s parents are away.46 We have here grandparents acting for their grand- daughters both in private and in public contexts. What is lacking, compared to cases with boys, are cases dealing with registrations of birth, apprenticeship, and epikrisis, which usually did not apply to girls; however, this can only partly explain the over-representation of boys, as these topics appear in only five cases from Oxyrhynchos.47 Why are girls not mentioned, then? One possible answer is hinted at in a letter from a grandmother in Alexandria to her daughter Sarapias: “I pray for your health, together with that of your household, and I make obeisance for you and for your child [masc.] in the presence of our lord Sarapis.” She goes on to write of the chiton which Sarapias has asked her to send “for the little one.” At the end of the letter, when the grandmother sends her greetings, we learn that Sarapias has other children too, who remain anonymous. It is possible that Sarapias herself and her son have been ill, and that this is why they are singled out for specific remembrance. However, as Roger Bagnall and Rafaella Cribiore point out, it might be more plausible to inter- 45 P.Fouad 35; P.Oxy. LXIX 3998; P.Mich. XVIII 789 (two girls); P.Giss. I 34; P.Oxy. III 496. 46 SB V 7558; SB V 7737; P.Brem. 63 (with P.Giss. 21, 23, 24, 77, and 78, all on Eudaimos and his granddaugher Heraidous; on this case see Hueb- ner, The Family in Roman Egypt 149); P.Col. VIII 215. 47 P.Oxy. LXVII 4585; P.Oxy. III 479; P.Oxy.Hels. 29; P.Oxy. LIV 3754; P.Oxy. XII 1452. VILLE VUOLANTO 391 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 pret this as an example of gendered bias: the grandmother would single out the person with the greatest potential to carry on the familial lineage, who was by default the oldest male grandchild. Other children are left without any specific men- tion, and we do not know if they were male or female.48 If this was the normal order of things, it could well have caused the bias we see in our material—indeed, the ordinary salutations by the grandparents, in which girls would be casually mentioned, also seem to be under-represented in these documents. More- over, males would inherit and thus own a greater part of the property than females—who held less than one-fourth of the landed wealth in the countryside, and far less in cities. Even if one might assume that the proportion for other property was higher, the educated guess that one-fifth of all property owner- ship would be in female hands could be too high.49 Thus, a greater proportion of the contracts dealing with property and inheritance would include boys rather than girls. 5. Action, interaction, and emotions: grandmothers and grandfathers What kind of influence did grandparents have on the lives of their grandchildren? We have already seen that, for infants, the grandmother’s role seems to have been important, through the support they would have given to the mothers while giving birth and immediately afterward. Moreover, women acted as guardians and supervisors for their grandchildren. But when a grandmother appears with her grandchildren, what kind of ex- pectations were there—was her role to be more passive, for example, than the role of a grandfather? In order to make such comparisons, it is necessary to limit our analysis to the material from Oxyrhynchos. Here we see that the roles which women took in the lives of their grand- children were indeed varied, and included representing them in public matters. For example, a grandmother named Aurelia 48 P.Mich. VIII 514, with R. S. Bagnall and R. Cribiore, Women’s Letters from Ancient Egypt (Ann Arbor 2006) 269–270. 49 R. Bagnall, Egypt in Late Antiquity (Princeton 1993) 92–99, 130–134. 392 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 Tayris wants to register a boy born to her son in the city archives. The boy’s father is alive: he is being reviewed for military service. In another document, a woman requests that a paternal grandson be registered into her house. In this case, too, it turns out that the father is alive, since he acts as the guardian for his mother in making the request.50 In both cases, the grandmother takes on the role of the family head in the context of an official transaction with the city magistrates. While it is unclear whether the boy actually moves to live with the grandmother in the first case, in the second there is no doubt about this—for some reason, the father can no longer take care of the boy, or the boy is moving in to help his grand- mother. In many cases, grandmothers are actively managing diverse family issues and helping their younger relatives. A maternal grandmother, Apollonous, is giving her daughter’s son as an apprentice to her son, a master weaver—who is the boy’s uncle. The grandmother seems to have lived with the boy—and she would do so also in the future, since it is agreed that Apol- lonous will feed and clothe him, although he is working with his uncle. The boy seems to be an orphan, and the grand- mother is taking care of the boy and of the continuity of the family trade.51 There is also a widow intervening on behalf of her daughter who had a 1½ year old nursling son, since her son-in-law provided no maintenance for his family; a grand- mother organizing a girl’s marriage together with the girl’s father; and a man arranging a visit of his child and his mother- in-law (that is, the child’s grandmother) to Alexandria to get medical help.52 On some occasions, the grandmothers are not 50 P.Oxy. LIV 3754 (the mother of the boy is not mentioned; the boy may still live with her); P.Oxy. III 479. 51 P.Oxy.Hels. 29, with V. Vuolanto, “Children and Work: Family Stra- tegies and Socialisation in the Roman and Late Antique Egypt,” in K. Mustakallio and J. Hanska (eds.), Agents and Objects: Children in Pre-Modern Europe (Rome 2015) 97–111, esp. 103. 52 P.Oxy. LIV 3770; P.Oxy. III 496; P.Oxy. XXXI 2601 (it is not clear who VILLE VUOLANTO 393 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 depicted as being in contact with their grandchildren, although they were contributing to the family group. An example is a case in which a wife asks her husband to ask his mother to take care of sending some items to her from the city of Oxyrhyn- chos—the couple has a son, who resides with the mother.53 In the Oxyrhynchos material, grandfathers appear in similar kinds of situations to grandmothers in both public and private contexts. They would have taken care of their young relatives if the parents were travelling or had died. In an epikrisis registra- tion of a 13-year-old boy by a family friend, the explanation for this action was that the orphan boy’s maternal grandfather had died quite recently; clearly the grandfather had taken care of the boy after the boy’s father had died. In another case, chil- dren would stay with their maternal grandfather and uncle while their mother was away. There is also a letter sent by a man, rejoicing that his daughter, her husband, and his grand- children are healthy, and caring about the financial interests of his granddaughter.54 In one will of a maternal grandfather, the grandson had already inherited an estate, while in another will, a grandfather includes his grandchildren in case his daughter were not still alive when he died.55 The more official role of grandfathers can be seen in the census declarations, in which paternal grandfathers register their grandchildren.56 A grand- father could also intervene on behalf of his grandchildren in matters of taxation, help his granddaughter in a lawsuit, or act to help a daughter as a kurios when she was applying for official guardians for her children (his grandchildren). And, as seen ___ had the eye problem in question). 53 P.Oxy. LXVI 3860. 54 P.Oxy. LXVII 4585; PSI XII 1247; P.Oxy. LIX 3998. For another letter in which a grandfather sends his greetings see P.Oxy. XII 1581. 55 P.Oxy. I 68; P.Oxy. I 105. 56 SB X 10220; P.Oxy. XXXIII 2671 and PSI I 53 col. i. Two other documents show us a grandfather publicly acting on behalf of his underage grandson: P.Oxy. I 54 and P.Oxy. VI 908. 394 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 above, they also served as official guardians for their underage relatives.57 Occasionally grandmothers and grandfathers appear to- gether, although this is rarer than one might expect. In a business letter, greetings are sent to the addressee’s wife, their sons, and grandchildren—who live in the same household or nearby. Grandparents are also mentioned in a case in which a guardian of their grandchildren returned their daughter’s dowry, since her husband had died.58 In a marriage contract, it was stipulated that the children would inherit from their ma- ternal grandparents if they died subsequent to the death of the children’s mother; here, it seems that the changing situation (the mother’s illness?) has prompted the contract, since it is stated that the couple has already been married for nine years, and have children.59 There is a certain expectation of the fragility of life—and a clear identification of oneself as a grand- parent, with duties towards posterity. Apart from general remarks about cohabitation and for- malised salutations, we have very little direct evidence about the emotional bond between the generations. For example, when a father writes to his daughter (and her husband), that he “rejoiced to hear of the health of you both and of your children,” although there is no specific reason to doubt the emotion behind the conventional formula, we can glean no concrete information about the emotional relationship between the man and his grandchildren.60 Or, how to assess the close- ness of the relationship between “little Pompeius” and his grandfather L. Pompeius Niger? In a letter, little Pompeius’ mother, Herennia, sends the boy’s greetings to her father, 57 P.Oxy. XIX 2235; P.Giss. I 34; P.Oxy. XXXIV 2709. 58 PSI III 236; P.Coll.Youtie 67. See also P.Oxy. VI 906, which documents a repayment of a dowry to a maternal grandfather in a case of divorce. 59 P.Oxy. XLIX 3491. 60 P.Oxy. LIX 3998. See also e.g. P.Oxy. XII 1582 (a grandfather sends greetings); BGU III 948 (grandchildren send greetings); SB V 7572 (a grand- son sends greetings). VILLE VUOLANTO 395 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 whereas in another letter she leaves three lines to be filled by the boy to greet his grandfather—but the plan came to nothing, and the letter we have has only the empty lines. In any case, Herennia was eager to promote bonding between the grand- father and the boy.61 More substantially, we have already encountered a grandmother praying for her grandchildren at Sarapis’ temple, and another trying to ensure that her grand- daughter would wear shoes because of scorpions. In one case there appears a boy, Ptollarion, quarreling with his grand- mother: does this kind of report in a private letter tell us about bad relations between the boy and his grandmother, or should this rather be seen as an indication of everyday interaction and nearness—sometimes inevitably leading to conflicts?62 In all, grandparents were active in the lives of their grand- children and their families, and contributed to the networks of everyday care by offering help and support on both personal and institutional levels. What is specifically noteworthy are the similar kinds of roles for both grandmothers and grandfathers: although more grandfathers than grandmothers appear, the more or less official tasks they took care of, and the ways they interacted with their grandchildren, are visibly similar. 6. Wider perspectives: grandmothers in the Roman Empire… The role of grandmothers—even in the presence of fathers, uncles, and stepfathers—was pronounced in the lives of many children. They acted as backups in case parents died or were for some other reason absent or temporarily unable to bring up their children.63 Grandmothers had authority in their families to oversee and protect their grandchildren and to make ar- 61 P.Mert. II 63 (SB VI 9122 = P.Oslo inv. 1444). The other letter is still unpublished; see D. Rathbone, “Pompeius and Herennia: A (Sad) Family Tale from Roman Egypt,” Omnibus 70 (2015) 19–20. 62 P.Mich. VIII 514; SB V 7737; P.Mil.Vogl. I 25. 63 See already Dixon, The Roman Mother 133, 154–155, 213; Cokayne, Experiencing Old Age 166; Harlow, in Age and Ageing 208; Parkin, in A Com- panion to Families 286. 396 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 rangements for their future. For grandparents in general, there was a special value in seeing their grandchildren grow up. Grandchildren meant the survival of the family and its tradi- tions, security in old age, and the maintenance and continuity of the names and memories of the older generations.64 According to legislation, the freedom of action for grand- mothers was rather narrow—narrower than that of any adult male relative. However, when a grandmother intervenes in the life of a grandchild, there is no difference between what she could do and what a grandfather could have done: grand- mothers and grandfathers, maternal and paternal, appear in all possible private, public, legal, and emotive contexts. The gendered bias visible in Roman literature, giving grandfathers more distant roles and instrumental attitudes towards their grandchildren (as mere propagators of their lineage),65 seems not to reach the everyday life of the more ordinary people. Moreover, as has been pointed out in studies dealing with the Roman aristocracy, there are no clear differences between ma- ternal and paternal grandparents in terms of taking care of the younger generation, even if, in legal contexts, the fact that a grandparent belonged to the maternal or paternal side would have made a difference.66 This may well be, as Tim Parkin proposes, because the legal definitions were “outweighed by feelings of, and moral obligations entailed by, pietas.”67 Un- fortunately, the papyrological material cannot give direct an- swers about the circumstances in Roman Egypt in this regard; but it is quite probable that demographic factors would have been important: quite apart from the family feelings or affec- 64 Vuolanto, Children and Asceticism 173–174, 186, 199, 202. 65 Cokayne, Experiencing Old Age 168; Dixon, The Roman Mother 33. 66 Parkin, Old Age 54–55; Krause, Witwen und Waisen III 62 and 73. See also Gardner, Family and Familia 244, on maternal and paternal relatives among the late Republican elites. Cf. Dixon, The Roman Mother 213, who highlights the importance of the women of the paternal side if parents had died. 67 Parkin, Old Age 56. VILLE VUOLANTO 397 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 tion, (physical) nearness and availability were the main reasons why grandparents intervened in children’s lives. Grandmothers were there, they were active, and they had everyday authority. In the quest for continuity of lineage and of the self, grand- children from the daughter’s side would, on average, have been available earlier than those continuing the male line, because of the differences between men and women with regard to the age of first marriage. There is one exception to this: maternal grandparents, especially grandmothers, appear especially in cases involving infants. This means that pregnant women, and those with a new-born child, would have been in frequent interaction with their own parents and would have sought support and comfort especially from their mothers. This ob- servation has also been made in earlier studies dealing with the Roman elites.68 Even here, however, one must bear in mind that most of these women actually lived with their mother-in- law and father-in-law, or with other parental kin. Maybe an indication of this kind of situation shows in a letter in which (probably) a maternal grandmother writes to her son-in-law that she “does not permit” her daughter to nurse the new-born child and tells him to hire a nurse instead. It would be com- pelling to see here a young mother seeking help from her parent—and the older generation, in turn, taking the oppor- tunity to further the sway of the maternal lineage in a situation in which the paternal line was in a much better position to in- tervene in the familial decisions of a young couple.69 If, qualitatively speaking, there were no clear differences in the roles taken by different types of grandparents, the preva- lence of paternal grandparents in quantitative terms needs 68 See Parkin, Old Age 55; Cokayne, Experiencing Old Age 166 (although I would be more cautious in seeing here affection a priori ); Dixon, The Roman Mother 94, 212, 217–218. 69 P.Lond. III 951 v. The writer of the letter seems to be a woman—in any case a maternal grandparent of the infant. See also e.g. P.Oxy. LIV 3770, in which a maternal grandmother intervenes on behalf of her daughter with an infant son. 398 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 explanation. It can at least partly be explained by the fact that quite often the wife had moved into the household of her husband’s parents: The paternal grandparents were more im- mediately involved in taking care of their grandchildren even when there were both paternal and maternal grandparents living. All other factors being equal, the paternal grandfather was the person to take care of the grandchildren. This prefer- ence in choice explains why, in spite of the demographic odds, grandfathers appear as often as grandmothers in the docu- ments. On the other hand, maternal grandmothers—even if they should, for demographic reasons, have been present more probably than any other group of grandparents—seem to have been underrepresented. There are no cases in which a grand- mother had taken an active public role with regard to her grandchildren in a situation when there appears also a grand- father. In short, the ideological preference for paternal and male grandparents seems to have outweighed any demographi- cally expected bias towards the maternal relatives.70 Another gendered bias in the sources concerns the absence of girls, at least partly because boys had a more central role in transferring the family patrimony forward. As in the case of paternal grandparents, here too we can see the result of the gendered expectations and the importance of the paternal lineage. On average, grandparents seem to have been more eager to express their worries about their grandsons’ wellbeing than about their granddaughters’. These results can be compared with research on grand- parents in other parts of the Roman world. Studies of com- memorative patterns outside of Roman Egypt have shown that grandparents were mentioned only in unusual cases where others were not available: if the parents had died, their children were more often commemorated by other persons (siblings, 70 This against e.g. V. Vuolanto, “Grandparents, Grandchildren,” in The Encyclopedia of Ancient History (Oxford 2012) 2978, claiming that “demo- graphic realities led to maternal grandparents taking on a greater role in the lives of their families than paternal grandparents.” VILLE VUOLANTO 399 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 friends, and dependants) rather than by grandparents. Accord- ing to Beryl Rawson, there are only 22 epitaphs from the city of Rome that mention a grandmother as a dedicator, 8 that mention a grandfather, and 2 that mention both. In Paul Gal- livan and Peter Wilkins’ study of family structures in Roman Italy, there appear around 1½ times more grandfathers than grandmothers, and 63% of these grandparents are paternal; in the case of epitaphs, one can note the overall low number of grandparents who are mentioned.71 These commemorative patterns seem to reflect not so much bad relations, or a lack of relations, but rather the demographi- cally determined odds against having long-lasting multigenera- tional households, and the fact that grandmothers were more probably at hand than grandfathers when their grandchildren died. On the other hand, cultural expectations highlighted the family nucleus as commemorators in inscriptions, and in literary sources there is a much higher visibility of males: these sources were produced by men who were interested in public life. In view of this, it is noteworthy that among the Roman elites of late antiquity, it was often the relationship with their grandmothers that the Christian writers, from Ausonius to Basil of Caesarea, wanted to highlight. Naturally, there were precedents for this in the New Testament.72 To scrutinize the cultural and social role of the grandmother among the Roman elites, however, would be a subject for another article.73 The very nature of the Egyptian material does not allow us 71 Rawson, Children and Childhood 241; Gallivan and Wilkins, in The Roman Family in Italy, esp. 268 with the table enumerating grandparents. See also Laes, Melita Classica 2 (2015) 104. 72 2 Tim 1:5; Ausonius Par. 5.9–10; Basil Ep. 204.6, 210.1, 223.3. See also e.g. Cic. Verr. 2.1.90–92; Sen. Ep. 94.5; Quint. Inst. 6 pr.7–9; Plin. Ep. 7.24.5; Tac. Ann. 4.22; Herodian 5.3.3, 5.7.5, 5.8.10; Jer. Adv.Ruf. 1.30; Greg. Naz. Ep. 160; Marc. Diac. V.Porph. 101. See further cases in Krause, Witwen und Waisen III 57–61 and 67–71. 73 Also in the Digest there is a rich array of cases involving grandparents, to be analysed in further studies. 400 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 to consider the importance of grandparents as ancestors in the lives of their grandchildren, in the sense of forming a base for family identity. For example, we do not encounter any family stories involving grandparents, nor do we hear of grandparents telling stories to their younger relatives—a theme well estab- lished in elite literature in both the early and the late Roman Empire.74 Another aspect in the relationships between children and grandparents which does not emerge in my sample of the papyri is the fact that some people depended on their grand- children in old age. However, this is only to be expected, since my material consists of cases involving underage children, who could hardly, by definition, assume such duties.75 Methodologically, the differences evident in my three sets of material, and in the epigraphic sources, illustrate well the dangers of using small data-sets—which is almost an inevitable situation in writing ancient history. Any interpretation has to be based on more than one data set, selected on independent criteria, and, as far as possible, to be backed with a qualitative analysis of the individual cases in order to provide insights into the family dynamics for the general interpretation process. Moreover, it is necessary to study male and female protagonists together; only thus can we contextualize the findings for a meaningful gender history. 7. …and beyond: the grandmother hypothesis I end my discussion by returning to the grandmother hypothesis. One has to draw a distinction between the bio- logical question of the possible link between evolutionary ad- vantages of grandmotherly care and the relatively long lifespan 74 Centlivres-Challet, in Families in the Roman and Late Antique World 9–13; F. Prescendi, “Children and the Trasmission of Religious Knowledge,” in Children, Memory, and Family Identity 76; Baroin, in Children, Memory, and Family Identity 42–43, 48; Parkin, Old Age 52. 75 Cocayne, Experiencing Old Age 168–170; Parkin, Old Age 53, 206–207, 210–212. See also Huebner, The Family in Roman Egypt 135–136, a case in which two sons and a grandson take care of a ‘retired’ couple in their old age. VILLE VUOLANTO 401 ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 after menopause, and the social and cultural advantages of different assistance strategies for grandparents. Roman Egypt cannot contribute much to the first question, but some obser- vations can be made about the second. Anthropological and demographic studies of various popula- tions have pointed out that, in fact, the effect of grandmothers is limited in most cases to the maternal grandmothers, whose presence indeed enhances the rates of fertility in the next gen- eration, and child survival among the generation following that. For paternal grandmothers the result is more ambiguous, even if frequently positive, while the presence of grandfathers seems not to have this effect.76 The material from Roman Egypt indicates that maternal grandmothers did indeed have a special place in the everyday life of their daughters and in con- tributing to the first years of their grandchildren’s life course. However, after their grandchildren’s infancy, maternal grand- mothers do not seem to have had any more special role in their lives than paternal grandmothers—and the roles taken by grandfathers and those of grandmothers were not dissimilar. In all, the evidence from Roman Egypt indicates that the scarcity of grandparents offers a plausible explanation of the similar roles that different types of grandparents would take later in the children’s life course: those grandparents who survived to see their grandchildren would contribute to their wellbeing. This result differs from many anthropologically- informed studies, which have revealed indifference towards their youngest relatives often even on the part of grandfathers and paternal grandmothers, whose presence may even have increased the mortality rates of their grandchildren.77 From the 76 The evidence is collected and compared with the effects of the presence of other relatives in R. Sear and R. Mace, “Who Keeps Children Alive? A Review of the Effects of Kin on Child Survival,” Evolution and Human Behavior 29 (2008) 1–18. 77 Sear and Mace, in Evolution and Human Behavior 29 (2008) 1–18; Sear, Allal, and Mace, Substitute Parents 66, 68–71, for rural Guinea. For historical examples see e.g. C. S. Jamison, L. L. Cornell, P. L. Jamison, and H. 402 GRANDMOTHERS IN ROMAN EGYPT ————— Greek, Roman, and Byzantine Studies 57 (2017) 372–402 point of view of the evolutionary aspects of the grandmother hypothesis, these first years of the human lifespan would be the most crucial, since the enormously high infant mortality rate formed an important threshold in the pre-modern world. On the other hand, from the children’s own perspective, and taking the grandchildren’s life course as a whole, the more equal distribution of involvement by paternal grandmothers and grandfathers later in childhood would have been more crucial for the childhood experience, and for the family dynam- ics. In Roman Egypt, the combination of virilocal marriage, a rather strong emphasis on male lineages, and demographic pat- terns seems to make the difference, highlighting the importance and equal role of all kinds of grandparents for the care, sup- port, and transfer of the intangible family heritage in childhood beyond infancy.78 December, 2016 University of Tampere History/Faculty of Social Sciences 33014 Univ. of Tampere, Finland Ville.Vuolanto@staff.uta.fi ___ Nakazato, “Are All Grandmothers Equal? A Review and a Preliminary Test of the ‘Grandmother Hypothesis’ in Tokugawa Japan,” American Journal of Physical Anthropology 119 (2002) 67–76, and E. Voland and J. Beise, “Oppo- site Effects of Maternal and Paternal Grandmothers on Infant Survival in Historical Krummhörn,” Behavioral Ecology and Sociobiology 52 (2002) 435– 443. As Sear and Mace (71) point out, given the huge variety of human populations, there is a danger in extrapolating any results. Still, in all tra- ditional societies the families seem to have benefited from the presence of at least one ‘extra’ (apart from the mother) relative, and “there is a variety of family structure in which children can thrive.” 78 I thank Tim Parkin, April Pudsey, Pia Vuolanto, and the anonymous referee for their most valuable comments and insight on this paper. I dedi- cate this article to my children’s grandmothers.