Tutela mulierum and the Augustan marriage laws*

The Augustan marriage laws have been the subject of extensive study and debate, in modern scholarship as in antiquity1. This paper does not seek to add to the vast bibliography setting out the terms of the laws, evaluating their purpose, or their demographic effect2. Instead, it addresses one specific question: how did the introduction of the ius liberorum (the “right of children”), which granted women freedom from tutela

(guardianship), affect the freedom of Roman women to deal with their property as they chose? This question -as much a matter of conditions prior to the Augustan legislation as the terms of the laws themselves -has produced diverging views: some scholars regard Augustus' laws as a watershed in the "emancipation" of Roman women, while others have stressed that republican women were already substantially emancipated and that the Augustan laws in effect gave legal form to existing social practice. As such, the question also has bearing on the larger question of tradition vs innovation, or anchoring innovation, in the Augustan principate -that is, how Augustus and his advisers could or did draw on existing values and legal/social realities to shape and support innovations 3 . In order to shed light on these questions, this paper seeks to reconstruct the operation of tutela mulierum, in its various forms, immediately before the passage of the Augustan laws. A further relevant factor, in assessing the impact of the ius liberorum, is to ask how many women were able to claim the privilege. Firm answers are not possible, due to the limitations of the evidence and uncertainty concerning the legal rules; however, I outline the parameters of the question, including the considerably higher bar set for Roman freedwomen. What emerges is that the reality of tutela and thus the potential benefit of the ius liberorum varied widely, depending on a woman's individual circumstances, the type of tutor she had, and especially whether she was freeborn or subject to the tutela of a male patron; moreover, while the Augustan laws paved the way for the further weakening of tutela for those freeborn women still subject to it, the gap between freeborn and freed grew wider. Irrespective of its practical effects, however, the ius liberorum possessed symbolic value for Roman women and for the social policy of Augustus and his successors, which may help to explain the retention of tutela mulierum of freeborn women as an institution even after reform under Claudius reduced it to little more than a formality.
The Augustan legislation on marriage comprised two laws -a lex Julia of (probably) 18 BCE and the lex Papia Poppaea of 9 CE. The two laws regulated marriage between different classes of person and introduced a complicated system of rewards and penalties designed to encourage marriage and the production of children, in large measure by modifying property rights 4 . So far as women were concerned, one important set of rules prevented unmarried persons from inheriting property from outside of close family, while those married but childless could take only half of any such bequest 5 . There were also limits on what a husband and wife could leave each other in their wills, unless they had three children, or a child in common 6 . Another rule stated that a woman who had three or more children (or four, in the case of freedwomen) was freed from tutela mulierum -the guardianship of women, or, more precisely, the need for an adult woman sui iuris to secure the authorisation of a male tutor for certain financial and legal transactions 7 . This privilege -the so-called ius (trium) liberorum 8 -constituted a significant legal innovation, in that, previously, the only women free from tutela were the Vestal Virgins, and Octavia and Livia by special grant in 35 BCE 9 . It also held considerable symbolic value, as an expression of an imperial social policy that prized motherhood, and as an honour system for Roman women 10 -even those who lacked the requisite number of children sometimes received the ius liberorum as an honour from the senate or emperor 11 . Its practical significance is debated, however: many scholars have emphasised that, by the late republic, the need for a tutor's authorisation was no more than a formality for many women, while others see the Augustan innovation as relieving women of a real burden 12 . Understanding the impact of the Augustan laws on women's financial freedom is therefore largely a question of understanding the operation of tutela immediately before the laws took effect.
In the Res Gestae, Augustus himself claimed that his new laws were anchored in -or rather "re-anchored" -ancient custom 13 , including the traditional value placed on marriage and raising children. Livy and 5 -Gaius 2.211, 286a. 6 -Tit. 194;Tit. Ulp. 11.28,29.3. See below on tutela. 8 -On this and other forms of "ius liberorum" under the laws, cf . Zabłocka 1988;McGinn 2013. 9 -Gaius 1.145; Dio 49.38.1 (see further below). Although all women were supposed to be subject to tutela (cf. Cic. Mur. 27), there were practical reasons why a woman might not have a tutor (see below on the lex Atilia).
10 -For instance, Propertius' representation of "Cornelia" (daughter of the future Augustus' second wife Scribonia) shows that the ius trium liberorum -and motherhood itself -conferred honour (Prop. 4.11 with Cairns 2006, 359); in Roman Egypt, evidence of women claiming the ius liberorum in contexts where it was not legally relevant suggests that it was a source of pride (Sijpesteijn 1965, 179). Cf. e.g. Dixon 1988, 89-90;Milnor 2008, 152-3;McGinn 2013;Armani 2018. 11 -Dio 55.2.6 states that honorific grants of the ius liberorum to men and women were made first by the senate and later by the emperor. Known recipients include such high-profile figures as Livia, the younger Pliny, and Suetonius, but also freedwomen (see e.g. Armani 2018 and below).
13 -RGDA 8.5; cf. Bellen 1987, 320-8;Morrell 2019, 13. Suetonius record that Augustus introduced the marriage legislation by quoting old republican speeches on similar themes 14 . Earlier legislation, as well, provided precedents or analogies for aspects of the Augustan laws 15 . At the same time, the laws broke with tradition in significant ways, both by intervening in private life as extensively as they did and on particular points; for instance, the pressure on widows to remarry within one year under the lex Julia of 18 BCE clashed with societal norms surrounding women's mourning and the ideal of the univira (the woman who had only one husband in her lifetime) and, probably for this reason, the rule was relaxed in the lex Papia Poppaea 16 . Indeed, social resistance more than once compelled the princeps to modify his plans 17 , while the laws he did pass had limited demographic effect (as the younger Pliny recognised) 18 . While Augustus took care to anchor the laws in terms of custom and the common good 19 , we have no statements from him or anyone else about the rationale for the creation of the ius liberorum -that is, the possibility of freedom from tutela -as a reward for motherhood 20 . The strategy might suggest that greater legal and financial freedom for women was something considered desirable (and socially palatable) at the end of the first century 21 . There were also precedents for freeing women from tutela in the Vestal Virgins (by the law of the XII Tables) 22 and the 14 -Liv. Per. 59;Suet. Aug. 89.2. 15 -Cf. Hopwood 2019, 75. Hopwood (2005 argues that the lex Voconia of 169 BCE, like the Augustan marriage laws, manipulated inheritance rules to encourage the production of children. Note also Caesar's lex Campana of 59 BCE, which distributed land to fathers of three or more children (Suet. Iul. 20 Raaflaub and Samons 1990. 18 -Plin. Pan. 26.5 (the laws were an incentive to rear children only for the rich). Tacitus (Ann. 3.28) considered the laws more effective at encouraging informers and enriching the treasury than increasing the birth-rate. See e.g. Brunt 1971, app. 9;Wallace-Hadrill 1981, 58-9;Treggiari 1991, 78-80;cf. Nörr 1981, 354, who suggests some contribution to population growth, especially outside Rome. We might say, therefore, that Augustus' "anchoring" efforts were only partly successful -or else that resistance to the laws was itself anchored in a longstanding tension between the value of children and the advantages of bachelorhood and limiting family size. Yet, insofar as the laws failed to increase the birth-rate, they served to enrich the treasury; see Wallace-Hadrill 1981, 72;Nörr 1981, 354-5. 19 -On the latter aspect, see Hopwood 2019, 72-5. 20 -Nor of community response to it, besides the fact that possession of the ius was regarded as an honour (see above). Eck (2019, 85) speculates that Roman men may have been displeased at losing control over women, but there is no evidence. We have no way of knowing if any woman ever sought to have children (so far as the matter was within her control) for the sake of gaining financial freedom; the law probably did have some impact on the family size of elite males: so Plin. Pan. 26.5; note also efforts to obtain the privileges of the law through adoption (Treggiari 1991, 79-80). 21 -Cf. Evans 1991, 15, who suggests that Augustus and his advisers identified and responded to a "real hindrance" for women in designing the reward under the laws.
22 -Gaius 1.145. Gaius states that the veteres wished the Vestals to be free from tutela on special grant to Octavia and Livia in 35, perhaps on analogy with the Vestals 23 . Unfortunately, we have only Cassius Dio's brief mention of the occasion, but possibly it was this moment, and not the lex Julia of 18, that represented the real innovation 24 -the moment when it became conceivable for women other than the Vestals (whose status was in many ways exceptional) to obtain complete legal freedom from male oversight of their financial affairs. "Complete legal freedom" is worth emphasising, however, because there were many property transactions for which a woman did not need her tutor's authorisation -and some evidence that women, at the end of the republic, were acting without authorisation where it was legally required. Moreover, the likelihood of a tutor obstructing a woman's wishes depended on a number of factors, especially the type of tutela she found herself in. As I outline below, this could make the difference between almost complete freedom for some women, while others might see their estates effectively frozen for the benefit of the tutors themselves -yet the distinction is sometimes obscured in modern accounts of the position of "women" at the end of the republic 25 . Another important variable is change over time -also sometimes obscured in scholarship on "Roman women" generally, and by the sources themselves, which are mostly late and reflect later legal change and juristic interpretation. When the jurist Gaius described the need for a woman to secure her tutor's auctoritas as an antiquated (and unwarranted) formality (1.190), he was writing in a legal context that had already seen the virtual nullification of tutela for freeborn women through praetorian intervention to compel tutors to give auctoritas in most circumstances and the abolition of agnatic tutela mulierum by Claudius. But the situation was rather different in 18 BCE. account of the honour of their office. However, modern scholarship has tended to emphasise the Vestals' anomalous legal position: as Gardner (1986, 24) puts it, because a Vestal left her birth family during her father's lifetime without emancipatio or passing into another family, "No one... could stand to her in the relation of tutor, whether legitimus, testamentary or any other sort". Cf. e.g. Zannini 1976, 23;Parker 2004, 574;DiLuzio 2016, 136-40. Sacchi 2003 and Gallia 2015 offer different interpretations.
23 -Dio 49.38.1; cf. Purcell 1986, 85;Cenerini 2016, 25-6. Octavia and Livia were also given tribunician sacrosanctitas. We do not know whether the privileges were conferred by lex, senatus consultum, or edict. It is tempting to speculate that the women (sister and wife, respectively, of the future Augustus) may have had some influence on the selection of freedom from tutela as a reward for mothers under the lex Julia. Livia herself had only two children but was granted the ius trium liberorum by the senate in 9 BCE, following the death of her younger son, Drusus (Dio 55.2.5-7).
24 -Cf. Purcell 1986, 85, who terms the grant "typical of the deviser in that, while traditional in flavour and nuance, in substance it was revolutionary and novel". 25 -Although, as Dixon (1984Dixon ( , 1984a and others have noted, republican trends saw an overall attenuation of tutela for freeborn women (as also manus marriage), its most restrictive forms remained alive and well into the Augustan age and beyond; see below on tutela legitima. Even if tutela was little more than a nuisance for most women, a significant minority remained for whom it was -at least potentially -a serious hindrance.
Attending to these changes is necessary in assessing not only the position of women and the impact of the ius liberorum, but also how the Augustan laws served to anchor later developments -or, rather, formed part of a continuum. What I offer here, therefore, is an overview of tutela mulierum in its various forms immediately prior to the Augustan laws (so far as it can be reconstructed), not to gainsay the general picture, but to highlight some of the nuances 26 .
As is well known, Roman women enjoyed relatively favourable property rights compared certainly to their counterparts in classical Greece and indeed in many much more recent societies 27 . An adult Roman woman sui iuris could own and control sometimes very substantial wealth. (Adult, for a Roman female, meant aged 12 or over; she was sui iuris if she was no longer subject to the potestas of her father, grandfather, or husband by manus marriage.) She could inherit property 28 ; she could also buy and sell property and engage in business (subject to the rules of tutela, as set out below). It was not unusual for a woman of senatorial or equestrian family to own multiple houses or farms. Terentia, the wife of the orator Cicero, is one well-known example 29 . In earlier times, the common practice of manus marriage meant that a woman and her property passed into the control of her husband 30 . However, by the late republic, at least, manus was uncommon 31 , and the usual form of marriage sine manu kept a wife's property rigorously separate from her husband's, to the point that even gifts between husband and wife were generally disallowed 32 . The only exception to this was the wife's dowry, which was legally her husband's for the duration of the marriage but had to be returned in the event of 26 -In what follows, I have not attempted to include full citation of sources and scholarship, especially where the legal rules are well established. Among modern studies, particularly useful discussions (with sources) include Watson 1967;Gardner 1986;Dixon 1984aand 1985. Zannini 1976and 1979and Medici 2013 offer extended studies of tutela mulierum. I have been somewhat more free than Watson in using later texts to elucidate republican law; in the case of tutela mulierum, there was a clear tendency towards relaxation of legal requirements, meaning that restrictions on women attested in a later period are, in general, likely to have applied in the late republic as well, and indeed many of the rules found in the legal sources are confirmed by republican evidence.
27 -On women in Greek law, see e.g. Sealey 1990, but cf. Blok 2018, arguing that Athenian women enjoyed greater freedom of economic action than traditional interpretations of the law have allowed. On the post-classical west, down to the twentieth century, see e.g. Dodds 1992, 902-4. Some countries today (notably Saudi Arabia) continue to practise male guardianship of women.
28 -See e.g. Crook 1986. A woman could be intestate heir to her near agnates; she could also inherit as heir or legatee under a will, subject to the restrictions of the lex Voconia from 169 BCE (references in Thomas 2007) and, later, the Augustan marriage laws. (On the relationship between the lex Voconia and the Augustan laws, see e.g. Spagnuolo Vigorita 2012).
29 -See Treggiari 2007, 34. Terentia's holdings included residential and agricultural properties, clearly more numerous than those that chance to be mentioned in the sources. 30 -Cic. Top. 23 suggests that, by his day, the property of a woman in manus was treated as dowry (cf. Dixon 1985, 165-6;Gardner 1986, 98).
31 -See e.g. Dixon 1985, 156-7. 32 -See Treggiari 1991 divorce or his death 33 . For the rest, the wife retained full ownership of her property and legally her husband had no say in how she used it. An illustration is Cicero's letter to Terentia of November 58 (Fam. 14.1.5 SB 8), protesting her decision to sell a row of houses during his exile: although Cicero objected, legally the property and the decision were hers alone. The only restriction on the woman's control over her property was that she needed the authorisation of her tutor (sometimes more than one) for certain important transactions. The institution of tutela mulierum was designed to protect the property of the agnatic family 34 : in its earliest form, by giving a woman's nearest male agnates (that is, relatives through the male line) the power to prevent her from disposing of property that would otherwise pass to the family on her death under the rules of intestate succession 35 . This power of tutela legitima (tutela arising by law) was extended to the patron of a freedwoman precisely because he was the woman's intestate heir 36 . While tutor is commonly translated as "guardian", the tutor of an adult woman did not administer her property or "look after" her in any other sense 37 . His only function was to give or withhold his auctoritas (authorisation) for particular legal and financial transactions, such as making a will, constituting a dowry, taking on obligations, entering manus marriage, manumitting a slave, and alienating (that is, selling or giving away) certain types of property known as res mancipi 38 -in essence, transactions that could diminish the woman's estate (though not all such transactions, as we shall see).
33 -Dixon 1984 discusses the Ciceronian (i.e. late republican) evidence. 34 -There is broad scholarly consensus on this point, which cannot be discussed in detail here; see e.g. Crook 1967, 113-14;Zannini 1976, 66;Dixon 1984a, 343 with n. 4;Medici 2013, ch. 3. The rationale found in (later) sources, which relates tutela mulierum to women's supposed weakness of mind (e.g. Cic. Mur. 27; Gaius 1.144, 190), seems to reflect later attempts (possibly influenced by Greek thought, and/or by analogy with the protective function of the tutela of minors) to explain an institution that had already lost much of its original relevance due to changes in inheritance patterns and women's economic engagement (see e.g. Schulz 1951, 180-91;Zannini 1976, 67-9;Dixon 1984a).
35 -Originally a woman could not make a will, meaning that her estate would pass to her agnates as intestate heirs (on intestate succession to women, see e.g. Crook 1986, 60). Later, as we shall see, a woman could only make a will with the auctoritas of her tutor and after breaking agnatic ties through capitis deminutio. A woman's tutores legitimi and her intestate heirs were identical, unless she had only female agnates in the nearest degree: Dig. 50.17.73.pr, from Q. Mucius Scaevola (cos. 95), with discussion in Watson 1967, 118-20. In the absence of agnates, both the hereditas and the tutela of a freeborn woman passed to the gens (see below).
36 -Gaius 1.165 (by analogy with the position of the agnates, who were both heirs and tutors). Conversely, the Vestal Virgins, lacking agnates, were also free from tutela (see n. 22). Tutela legitima and other categories of tutela mulierum are discussed further below. 37 -Cf. Gaius 1.190. By contrast, the tutor impuberis did administer (Tit. Ulp. 11.25). In this paper, I use tutor, without italics, to denote the tutor mulieris.
Whether a woman also required her tutor's authorisation to purchase res mancipi is less clear. Gaius (2.80) and the Tituli of Ulpian (11.27) specify that she needed auctoritas to alienate, and Gaius also states that anything, whether res mancipi or not, can be paid over (solui possunt) to a woman without auctoritas (for instance, in release of a debt), because women are allowed to improve their position even without auctoritas (2.83). On the other hand, Tit. Ulp. 11.27 states that a woman needed auctoritas to undertake civil business (civile negotium), which could include mancipatio (the procedure required for formal conveyance of res mancipi) 39 , and Apuleius refers to his wife Pudentilla having her tutor's authorisation for the purchase of a farm 40 . Further, Vat. fr. 45 indicates that a woman needed auctoritas for in iure cessio (transfer by means of a fictitious trial) of res nec mancipi. However, even if a woman required auctoritas to make a formal purchase of res mancipi, she could still acquire full ownership through an informal purchase followed by usucapio (acquisition of ownership by possession, for one year for movable property or two years for immovables) 41 . It seems possible for this reason, and also perhaps because the acquisition of res mancipi would tend to improve the estate 42 , that auctoritas was not required -or came not to be requiredfor the purchase of res mancipi.
Res mancipi included land in Italy, slaves, and beasts of burden (Gaius 2.14a). All other things were res nec mancipi, and these a woman was free to dispose of as she pleased, without her tutor's authorisation 43 . Notably, res nec mancipi included gold and silver and provincial land (Gaius 2.20-1): potentially very valuable property 44 . We know of at least one republi-39 -The rules about alienation and about civil law transactions appear frequently in scholarship, with little attempt to reconcile them. Zannini (1976, 100-1), however, argues that the specification of alienation would be redundant if civile negotium is taken to include all civil law transactions (as some scholars have held), which would also conflict with Gaius' testimony that women administer their own affairs (1.190) and can make loans without auctoritas (2.81). Zannini suggests that Tit. Ulp. 11.27 should be taken as referring to civil business that placed the woman under an obligation. 40 -Apul. Apol. 101. The farm in question was probably in Africa. Only Italian land was res mancipi, but Pudentilla's farm may have been in a colony with the ius Italicum: Evans Grubbs 2002, 34. 41 -On usucapio, see e.g. Buckland and Stein 1966, 242. One of its chief functions was to give dominium where res mancipi was transferred by mere traditio (delivery). In the interim, the purchaser had bonitary ownership, which protected against even a claim by the dominus (191-2). See below on the question of usucapio from women in tutela. Jakab (2013, 148-9) suggests that women could circumvent the need for authorisation by conducting business through a male agent, but this seems not to have been not possible for mancipatio (Buckland and Stein 1966, 277-8).
42 -Cf. Gaius 2.83. 43 -E.g. Vat. fr. 259 (Papinian) describes the case of a woman who made a gift of provincial land, slaves, and livestock to an Italian man without her tutor's authorisation; the gift was effective immediately with respect to the land (provincial land being res nec mancipi), but not the slaves and animals (res mancipi), which also could not be usucapted due to the lack of auctoritas.
44 -Res mancipi corresponded to the most important property in Rome's archaic agrarian society; the key distinction, however, was not one of value but rather of method of conveyance: res can woman -Cicero's friend Caerellia -who owned substantial provincial estates (Cic. Fam. 13.72 SB 300), while women's jewellery (also res nec mancipi) was at various times a target for censors and triumvirs and a war chest for Rome 45 . A woman could also make a loan without auctoritas (and thus place others under obligation to her), as money was res nec mancipi (Gaius 2.81).
While a woman could do whatever she liked with her res nec mancipi during her lifetime, however, in order to make a valid will she required her tutor's authorisation 46 . Furthermore, a freeborn woman (other than a Vestal) could not make a will until she had undergone capitis deminutio (change of legal status), which served to break agnatic ties 47 . For women who had not previously done so (for instance, by entering manus marriage), this was accomplished through coemptio, a form of fictional sale, which required the auctoritas of her existing tutor(s) and also resulted in a change of tutor (Gaius 1.114-15). Freeborn women thus required auctoritas twice: first for undergoing coemptio and again for making the will itself. The latter was likely often a formality, since a woman undergoing coemptio could be expected to choose some pliable person to serve as her new tutor 48 , and because capitis deminutio severed agnatic ties, meaning that an agnatic tutor had already forfeited any intestate claim on the woman's estate if he authorised coemptio. For that reason, however, auctoritas for coemptio may have been hard to come by, especially for women in agnatic tutela 49 . A freedwoman in the tutela of her patronus was in a similar position: she did not undergo coemptio or change of tutor 50 , but mancipi had to be transferred by mancipatio (or in iure cessio), whereas ownership of res nec mancipi passed on simple traditio (Gaius 2.18-22).
45 -E.g. Liv. 34.5.9-10, 39.44.2; App. BC 4.33. 46 -Roman women were originally unable to make wills, but, at some time after the XII Tables, acquired the capacity to make wills in the form per aes et libram. By Polybius' day, at least, it was common for wealthy women to make wills (see Dixon 1985, 170).
48 -Coemptio was also employed where a woman wished to exchange her existing tutor(s) for one of her choice (see below). It seems possible, however, that some tutors might have given or withheld auctoritas for coemptio depending on the identity of the intended new tutor (for instance, another relative), with the intent of retaining some control over the woman's property.
49 -Cf. Medici 2013, 168-70. 50 -Watson (1967 argues that the coemptio requirement applied to all women, freedwomen included, but Gaius 3.43 indicates that a freedwoman's patronus gave auctoritas for the actual making of the will, which suggests that she had not undergone coemptio (even if she had been remancipated back to her original patron, his standing would then be that of tutor fiduciarius and no longer tutor legitimus). It is also worth noting that Livy's story of Hispala Faecenia requesting a tutor and making a will (39.9.7) seems to leave no room for a change of tutor by coemptio, although, as Watson notes, this evidence should not be pressed. Possibly manumission counted as sufficient change of status (cf. Gardner 1986, 167, although the manumissio of a slave was not technically capitis deminutio: required her patron's auctoritas to make a will -and a patron could be expected to protect his interests by withholding it (Gaius 3.43).
In some circumstances, already in Cicero's lifetime, the praetor would give effect to a will that was invalid under civil law by granting bonorum possessio secundum tabulas 51 . It is not clear if this was available for wills made without authorisation, but bonorum possessio was excluded where a woman had not undergone capitis deminutio (Cic. Top. 18) and was of no avail against intestate heirs (Gaius 2.119); thus a woman could not use an unauthorised will to defeat the claims of her agnates or the patronus of a freedwoman 52 . For these reasons, the need for auctoritas could represent a serious limitation on a woman's freedom to dispose of her property, especially for those in tutela legitima (see further below).
How likely a tutor was to give his consent to a particular transaction would depend on what the woman wanted to do (I will return to this shortly) and individual inclination; he also needed to be physically present at the relevant time, which often created difficulties, to judge from later legal rules 53 . But the most important variable was the type of tutor a woman had. The key distinction was between what we might call interested and disinterested tutors: that is, those who could expect to benefit from the woman's estate if she died intestate, and therefore had an interest in preventing her from making a will, alienating valuable res mancipi, or taking on obligations that might diminish her estate, and those with no such vested interests 54 . To the first category belong the various forms of tutela legitima, where a woman fell by law into the guardianship of her nearest agnates (or the gens, in the absence of agnates); her male patron (or patron's son), in the case of freedwomen; or her father, in the case of manumitted daughters; as well as agnate relatives appointed by will. The second comprised tutors external to the family chosen by the Dig. 4.5.3.1), or the difference may be explained by the fact that freedwomen lacked agnates (so e.g. Zannini 1976, 166;Dixon 1984a, 346 n. 14). The rule is in any case consistent with other rules that tended to keep freedwomen under the power of their male patrons (see Perry 2014, ch. 3 and below). 51 -As in the case of Turpilia's will (Cic. Fam. 7.21 SB 332, June 44). The legal details were, and remain, disputed; Watson (1971, 73-5), followed by Shackleton Bailey (1977, 2.472-3), holds that Turpilia had not undergone coemptio and therefore lacked testamenti factio (capacity to make a will at all; cf. Cic. Top. 18, 50); Tellegen-Couperus and Tellegen (2006, 387-8) argue that she acted without a guardian. Cf. Crook 1986, 72.
52 -This was surely one of the core functions of tutela (cf. e.g. Zulueta 1953, 51). Antoninus Pius modified the situation for wills of women not in tutela legitima (Gaius 2.120-2). 53 -Gaius 1.173-6; Tit. Ulp. 11.22: a senatus consultum of unknown date (but evidently later than the lex Julia: Gaius 1.177-8) provided for the appointment of a replacement tutor, in many circumstances, where the existing tutor was absent.
54 -The distinction is mine; Roman taxonomies of tutela varied (Gaius 1.188). The two categories align approximately with the tutor legitimus and the so-called tutor extraneus (external to the family), respectively; however, it was possible for a woman to have an "interested" agnate as tutor who was not tutor legitimus (see below). woman's father (or husband by manus) in his will or, in some circumstances, by the woman herself.
One common way in which a tutor was appointed was by a woman's father in his will, as also for any prepubescent children 55 . This is known as testamentary tutela. We have little information about the tutors chosen for women, but relatives, family friends, and freedmen -or some combination of these -were probably usual 56 . Testamentary tutela was at least as old as the XII Tables 57 and modern scholars sometimes treat it as the norm 58 , but caution is warranted. Even in the late republic, significant numbers of women continued to fall under the tutela legitima of their agnates (see below). Furthermore, a father could choose one or more agnatic relatives as testamentary tutor(s); an agnate appointed in this way would be less strongly protected than the tutor legitimus but still able to safeguard his own interest in the woman's property by withholding auctoritas and, in particular, by preventing her from making a will 59 . Where a manus marriage existed (which was rare, but still attested at the end of the republic) 60 , the husband could appoint a tutor by will for his widow (Gaius 1.148). Alternatively, he could give his wife the right to choose her own tutor (so-called optio tutoris), either once or multiple times 61 . If the husband made no provision in his will, his wife would fall into the tutela legitima of her agnates, which meant, for a wife in manus, her husband's agnates -including, potentially, her own sons 62 .
Optio tutoris does not seem to have been available in the case of fathers and daughters. However, as already mentioned in connection with wills, it was possible for a woman to change her tutor through coemptio (imaginary sale). This required the authorisation of her existing tutor (or tutors) 63 , 55 -Gaius 1.144, 149. 56 -That is, the same sorts of people as were usually chosen for children (cf. Gardner 1998, 242). Tutors were appointed for adult daughters in the same way as for minors (Gaius 1.144-5). Indeed, Gaius' statement that a daughter remains in tutela after puberty (in tutela permanet, 1.145) suggests that she would retain the same tutor from childhood into adulthood (provided, of course, that he did not die in the meantime); the reference to tutori a pupillatu in CIL 6.2210 may be an example of this. Cf. Medici 2013, 17. 57 -Tab. V.6 prescribed that, where there was no tutor (either because there was no [valid] will, or no tutor appointed), tutela went to the nearest agnate: Gaius 1.155 with Crawford 1996, 642. 58 -E.g. Gardner 1986, 15;Tellegen 2013, 408-9. 59 -Later, the praetor began to compel tutors other than tutores legitimi to give auctoritas, but this was most likely a post-republican development; see below on this point and on the special privileges of the tutor legitimus.
63 -Gaius 1.115. A tutor legitimus could also choose to transfer the tutela to another person (1.168-70). but, provided he consented, the woman was free to choose some biddable person -even her own freedman 64 , for example -as her new tutor (termed tutor fiduciarius). Cicero may allude to this device in Pro Murena when he says that lawyers have found ways to subject tutors to the power of women, and not the other way around 65 .
A further possible way in which a woman could choose her own tutor was where a tutor was appointed by a magistrate. For women in Rome, the lex Atilia provided that a woman who had no tutor could have one appointed by the praetor and a majority of the tribunes of the plebs 66 . A similar process applied for Roman women in the provinces under the lex Julia et Titia, usually identified as a law of the future Augustus and his colleague in 31 BCE 67 . A woman might have no tutor where she had no testamentary tutor (or her tutor had died) and no relative eligible to be tutor legitimus, or, in the case of freedwomen, where her former owner was a woman (and therefore could not be tutor: Gaius 1.195). The lex Atilia is generally thought to have been passed around 210 BCE, at a time when the Punic wars had left many women without fathers, husbands, or tutors 68 . It was the woman herself who requested the appointment of a tutor 69 ; to judge from clauses in later municipal laws which probably derive from the lex Atilia, she also nominated the person she wanted appointed as her tutor 70 . For this reason, and also because a woman with no tutor ipso facto had no agnates or patronus with an interest in her estate, the law should perhaps be seen as facilitating rather than controlling 64 -CIL 6.7468 is a possible example of this: Dixon 1984a, 347 n. 18. 65 -Cic. Mur. 27 (hi invenerunt genera tutorum quae potestate mulierum continerentur); see e.g. Dixon 1984a, 347. 66 -Gaius 1.185 (cf. 1.187 on guardians taken captive); Tit. Ulp. 11.18. Besides tutors assigned under the lex Atilia, a substitute tutor could be appointed by a praetor where there were legal proceedings between a woman and her tutor (Gaius 1.184; cf. Watson 1967, 128). 67 -Gaius 1.195. Magistrates of Italian towns may already have had the power to appoint guardians (Crawford 1996, 447); cf. n. 70, below, on the lex Ursonensis (Caesarian in date, but with later insertions: Crawford 1996, 395-6).
69 -Gaius 1.185 and Tit. Ulp. 11.18 state that tutors are appointed (datur; dari) by the praetor and tribunes. These passages tell us nothing of how the process was initiated. However, Liv. 39.9.7 states that Faecenia received a tutor after petitioning the praetor and tribunes (tutore ab tribunis et praetore petito), and Gaius 1.195b specifies that a freedwoman had to request a tutor (tutor peti debet) if her patron was adopted. Cf. next note.
70 -E.g. Lex Irnitana ch. 29, ll. 18-19 provides that the woman herself, who lacks a guardian or whose guardian is uncertain, should request that the duumvir appoint a guardian and nominate the person she wants appointed (postu[l]averit, uti sibi tutorem det, <et> eum quem | dari volet nominaverit); lex Ursonensis ch. 109, though highly fragmentary, seems to have contained a clause to the same effect. On the relationship between these laws and the lex Atilia (and lex Julia et Titia), see e.g. Watson 1967, 124;Nörr (2001) and Galaboff (2016, ch. 3) go further in attempting to reconstruct clauses of the earlier statutes. women in dealing with their property -that is, by enabling women who otherwise lacked tutors to engage in property transactions that required tutoris auctoritas 71 .
By the late republic, then, it was very possible for a woman to have a tutor who would act as no more than a rubber-stamp for her business affairs. Scholars regularly observe that many women were able to do as they pleased, unhampered by tutela 72 ; to take the example of Terentia again, we hear quite a lot about her business affairs, but her tutor is never even named 73 . However, we should not assume that all women were in the same position 74 .
Although fathers could appoint tutors for their daughters by will, some apparently chose to leave their daughters to the tutela legitima of agnate relatives: late republican jurists discussed cases where fathers appointed tutors for sons but not daughters 75 . In addition, tutela legitima could arise where a father died intestate, or his will failed, or (assuming the Digest reflects republican law on this point) if the testamentary tutor died 76 . Tutela would then fall by law to her nearest male agnate or agnates, such as her brothers or paternal uncles (Gaius 1.155-6). All agnates in the nearest degree became tutors. A probable example is Clodia: Cicero (Cael. 68) implies that she was in the tutela of her brothers (on whose auctoritas she manumitted some slaves); the casual reference suggests, furthermore, that her situation was unremarkable -and, in Clodia's case, apparently little hindrance to her business activities 77 . 71 -Cf. e.g. Gardner 1993, 95;Nörr 2001, 69;Medici 2013, 131-3. By contrast, Evans (1991 characterises the lex Atilia as a governmental response to what was seen as excessive female freedom and Webb (2014, 70-5) goes so far as to think of "state tutelage" aimed at controlling women's property.
72 -E.g. Weiand 1917, 383;Pomeroy 1975, 151;Dixon 1984a, 347;Dodds 1992 (cos. 51) for the view that the appointment of tutors "filio filiisque" applied only to sons, contrary to the usual rule that the masculine includes the feminine; cf. Watson 1967, 115-16 and1991, 35-6. The passage suggests, further, that the will in question was challenged, perhaps by the testator's daughter(s) or the would-be tutors. Cf. the Laudatio Turiae, cited below, for an attempt to invalidate a father's will and claim tutela legitima over his daughter. 76 -Dig. 26.2.11.3 (Ulpian), 26.4.6 (Paul). The same demographic factors that meant many women lost their fathers by puberty and most by age 25 (Saller 2007, 91) meant also that any tutor appointed by a woman's father was likely to predecease her. It is not clear if it was possible, during the republic, to appoint successive testamentary tutors by will; see Watson 1967, 117. 77 -Cic. Cael. 68: At sunt servi illi de cognatorum sententia, nobilissimorum et clarissimorum hominum, manu missi. Tandem aliquid invenimus quod ista mulier de suorum propinquorum, fortissimorum virorum, sententia atque auctoritate fecisse dicatur. ("'But these slaves have been given their freedom on the advice of the woman's relations, illustrious men of the very highest rank'. At last we have found something that she is supposed to have done on the advice and authority of her valiant If there were no male agnates, then tutela could pass to the gens (the larger family unit). This remained a real possibility at the end of the republic, as is clear from the Laudatio Turiae, where the husband praises his wife for defeating an attempt by false kinsmen to invalidate her father's will and thereby become her tutors, presumably with the intention of inheriting her estate 78 . A passage from Catullus plays on what must have been a familiar fear of tutela legitima (in this case, again, the tutela of the gens) as an opportunity for greedy relatives to get their hands on a woman's property: Catullus 68b, ll. 119-24 describes the joy of an old man on the birth of a son to his only daughter, thus enabling him to leave his estate to his grandson and exclude the gentilis who would have become his daughter's tutor (and thus heir) if he died intestate 79 . The tutela of the gens could arise only in the absence of both testamentary and agnatic tutors, yet Catullus' allusion to it, without explaining the legal situation, suggests that it was familiar to late republican readers. Agnatic tutela must have been far more familiar, and perhaps reasonably common, before its abolition by Claudius 80 .
As for freedwomen, a female slave freed by a male owner automatically fell under the tutela of her patron, or his son (when the patron died), even if the son was a minor and therefore unable to exercise the functions of tutor 81 . The Augustan marriage legislation addressed this situation to some extent by providing that a freedwoman in the tutela of an impubes could have an alternative tutor appointed for the purpose of constituting a dowry (Gaius 1.178), but the general rule is testament to the privileged position of the patronus vis-à-vis his freedwomen 82 . By contrast, as I have noted, a woman freed by a female owner did not have a tutor legitimus and instead could request a tutor under the lex Atilia (Gaius 1.195 Watson 1967, 121-2;Osgood 2014, 20-4. 79 -See e.g. Gardner 1986, 193;Osgood 2014, 21. Catullus here assumes the reader's familiarity with both the law of tutela and the lex Voconia: the latter prescribed that a woman in the first property class could take only half of an estate in a will, meaning that the father could not appoint his daughter as his heir. If he died intestate, the daughter could take the whole estate, but would fall under a tutor legitimus, who could prevent her from making a will and thus become her heir on intestacy. The birth of a grandson (that is, a male heir) enabled the grandfather to make a will benefiting his descendants and exclude the gens. 80 -Cf. e.g. Watson 1967, 148;Evans Grubbs 2002, 24;Levick 2015, 146. For late republican examples, see n. 77, above. Gaius states that agnatic tutela of women was abolished by Claudius (1.157), but nonetheless describes many of the rules pertaining to it (e.g. 1.155-7, 1.168, 2.47). By contrast, the tutela of the gens fell out of use earlier (cf. 3.17) and Gaius does not describe it, yet even this form seems to have been familiar at the end of the republic. On the related question of the prevalence of intestacy, see e.g. Crook 1986; Gardner 2011. 81 -Gaius 1.165, 175, 178. Watson (1967 with n. 1) makes the case that minors could be tutors only in the case of freedwomen; in other cases, tutors had to be puberes.
82 -See e.g. Watson 1967, 118 n. 1;Perry 2014, esp. 84. The chief reason why tutela legitima was particularly burdensome was because the tutor could quite legitimately use his position to prevent a woman from making a will or alienating res mancipi, in order to keep her property for himself or for the agnatic family. Roman law frankly acknowledged and indeed protected this right: for instance, Gaius (2.43) attests the expectation antiquo iure (prior to the Augustan laws) that a patron would safeguard his interests by preventing his freedwoman from making a will, and had only himself to blame if she made a will in which he was not named heir 83 . Later, when it became possible to compel tutors to give auctoritas, exception was made for the tutor legitimus, for his own benefit -specifically, to preserve the property he could expect to inherit as intestate heir 84 . Significantly, the tutor legitimus would exclude even the woman's own children if she died intestate, because in Roman law the relationship between mother and child was cognatic, and both agnates and patrons (as legitimi) ranked above cognates in the order of succession 85 . This was despite the social expectation that women would leave their property to their children: this is reflected, for example, in the Laudatio Murdiae (of Augustan date, or thereabouts), where a son praises his mother particularly for the equitable manner in which she divided her property among her children from two marriages 86 . That is, in order even to leave her property to her children, a woman had to make a will, but she could not make a valid will if her tutor refused to give his authorisationand a tutor legitimus had a vested interest in refusing.
Besides this, stricter rules applied to tutela legitima than to other types of tutors; rules which tended to preserve the estate for the intestate heirs 87 . For instance, a woman in tutela legitima had to secure the auctoritas of all her tutors, if she had more than one, whereas the consent of one tutor was sufficient for testamentary tutors 88 . Another rule, attributed to 83 -Cf. 1.165, mentioned earlier, where Gaius explains that the tutela of patrons was derived from their inheritance rights.
84 -Gaius 1.192. The passage refers specifically to patrons and fathers of manumitted daughters -by that time (after the abolition of agnatic tutela: 1.157) the only remaining categories of tutores legitimi for adult women. They could be compelled only in exceptional circumstances.
85 -See e.g. Gardner 1986, 190-3;Watson 1971, 176-8, 183-4. Succession to cognates seems to have been introduced in the lifetime of Trebatius Testa, so perhaps at the end of the republic; they ranked after a woman's legitimi. In civil law, since a woman did not have sui heredes, her intestate heirs were her agnates, or, failing agnates, the gens. Only after the senatus consultum Orphitianum (178 CE) were a woman's children able to inherit from her in the first line, as legitimi.
86 -Laudatio Murdiae (CIL 6.10230), ll. 4-5; see Lindsay 2004, 94-6. Cf. e.g. Val. Max. 7.7.4;Cic. Fam. 14.1.5 SB 8. 87 -Similarly, when it became possible to compel a tutor to give auctoritas, exceptions were made for tutores legitimi (Gaius 1.190,192). We do not know when tutors became compellable, but there is no evidence for the republic. A freedwoman in tutela legitima also could not have a replacement tutor appointed if her patron was absent, except in special circumstances (Gaius 1.174, 176-8). Tit. Ulp. 11.26. the XII Tables, provided that the property of a woman in tutela legitima could not be usucapted unless it had been transferred with the authority of her tutor 89 . Among other implications, for women in tutela legitima this closed off what was potentially one way of circumventing the need for a tutor's authorisation to alienate res mancipi: by making an informal transfer which would then give the possessor full ownership after one or two years. It is not clear, however, if a purchaser could usucapt res mancipi transferred by a woman without her tutor's authorisation if he knew she acted without auctoritas (or with the authorisation of a false tutor) 90 . This raises the question of law vs practice. Some scholars have argued that women in the late republic were regularly circumventing the rules of tutela. Such a gap between law and practice would have implications not only for women's legal and financial freedoms, but also for how we evaluate the Augustan laws: for instance, Jakab argues that the widespread practice of doing business through slaves and freedmen allowed women to sidestep the legal requirement of a tutor's authorisation 91 , and that Augustus, by legislating greater liberties for women, was recognising earlier social and economic developments 92 . Tellegen-Couperus goes further, contending that women at the end of the republic might quite commonly sell res mancipi without a tutor's authorisation and have their actions upheld by the praetor 93 , and that there was at least a school of thought, 89 -Cic. Att. 1.5.6 SB 1; Gaius 2.47. Neither could manus marriage arise by usus for a woman in tutela legitima, except with the auctoritas of all her tutors: Cic. Flacc. 84; Watson 1967, 21-2. 90 -Usucapio required bona fides on the part of the possessor. Vat. fr. 1 ascribes to the veteres (pre-Labeo, thus probably republican) the view that "Someone who knowingly buys res mancipi (from a woman) without her tutor's authorization or with the authorization of a false tutor who he knew was not (her real tutor), appears not to have bought in good faith" ([Qui a muliere] sine tutoris auctoritate sciens rem mancipi emit vel falso tutore auctore quem sciit non esse, non videtur bona fide emisse, trans. Evans Grubbs). The passage goes on to say (after describing various juristic opinions) that Julian, propter Rutilianam constitutionem, allowed the purchaser to usucapt if he had paid the purchase price, i.e. whether or not he knew that the woman had acted without auctoritas (but the woman could stop usucapio by returning the money). The "Rutiliana constitutio" is sometimes assigned to P. Rutilius Rufus (cos. 105), but it is not clear that this was a republican rule and Bauer (1986, 99) argues that it did not exist at all (she emends propter utilitatem constituit). It is also unclear how Vat. fr. 1 should be reconciled with Gaius 2.47, which implies that the property of women not in tutela legitima could be usucapted (of course, that need not be by a purchaser, knowing or otherwise). These problems cannot be considered in detail here; for discussion, see e.g. Hausmaninger 1964, 13-27, 39;Tellegen-Couperus 2006, 427-9;Pool 2016, 80-2. What is reasonably clear is that a purchaser could acquire ownership by sale plus usucapio from a woman not in tutela legitima if he did not know that she acted with a falsus tutor, and also that some women, already in the republic, attempted to circumvent the requirement of auctoritas by this device. Ignorance or mistake of fact as to whether or not a woman was in tutela was only likely to occur after the Augustan laws; cf. Hausmaninger 1964, 23-4. 91 -Jakab 2013. Her evidence comes from the mid-first century CE "Archive of the Sulpicii", but she suggests that it reflects republican practice (125). Gardner 1999 offers a different interpretation of the evidence.
92 -Jakab 2013, 125. Cf. e.g. McGinn 2003, 82. 93 -Tellegen-Couperus 2006. However, her argument, based on Vat. fr. 1, seems to misread Labeo's opinion (compare Hausmaninger 1964, 19), and certainly goes too far in concluding, in the mid-first century BCE, that would allow a woman to make a will without auctoritas 94 . What seems clear is that lawyers at the end of the republic were dealing with situations where women acted without their tutors' authority 95 . There will also have been many cases that "flew below the radar", especially for smaller transactions or where there were no rival interests involved; in general, it was wealthy women who had the most at stake and who were also the most likely to be held to the legal rules. However, it is probably safer to see such cases as evidence of frustration with the constraints of tutela rather than accepted practice -certainly the use of a false tutor (Vat. fr. 1) speaks against any generally accepted view that a woman was free to act without auctoritas. In other words, it appears that, at the end of the republic, there was a "market" for legal changes that would free women from tutela.
Beyond the legal rules, the lived reality of tutela would vary also depending on the personality of the tutor and what exactly the woman wanted to do. Not all tutors were self-serving 96 . It is clear that some tutores legitimi considered their role a burden rather than a boon 97 . Later epigraphic evidence shows that there could be trusting, even affectionate relationships between women and their tutors 98 . Moreover, even a domineering tutor legitimus might have supported a canny businesswoman in managing her property portfolio 99 . On the other hand, he might have taken a very different attitude if she proposed to donate valuable land for the benefit of her city or leave her estate to her favourite mime actor -transactions that would transfer property out of the familia 100 . It is plausible, therefore, from this evidence, that the praetor probably authorised usucapio in all cases where res mancipi was transferred by a woman without auctoritas, and whether she was in tutela legitima or otherwise (428,434). She may nonetheless be right that praetorian law was more permissive than the civil law rules found in Gaius.
96 -As Gardner (1986, 22) notes, we should not assume that even a tutor legitimus normally withheld his auctoritas, simply because he could.
97 -Gaius 1.168. The tutor of a woman could transfer his duties to another person by in iure cessio; by contrast, the tutela of a male minor could not be transferred, because it ended at puberty and therefore was not considered onerous. Cf. Dixon 1984a, 353-4. 98 -E.g. Claudia Quinta provided a tomb for C. Julius Hymetus, whom she describes as her child-minder, guide, and tutor from girlhood(?) (paedagogo suo καὶ καθηγητῇ item tutori a pupillatu, CIL 6.2210); Furfania Saturnina (CIL 6.2650) and Ostoria Acte (CIL 6.7468) included their tutors in their family tombs. 99 -Clodia may be an example of a woman in tutela legitima allowed to do as she pleased (see Dixon 1984, 347 and above). Cf. Gaius' comment that women are able to improve their position without auctoritas (2.83).
100 -Cf. the case of Gegania, who became notorious by making her slave Clesippus (purchased as a "package deal" with a costly lamp stand) her lover and her heir cf. CIL 10.6488). Herrmann (1964, 109-10) notes that Gegania cannot have had an agnatic tutor, for he surely would not have approved her intentions. that the Augustan laws were a factor in the emergence of women as civic patrons and benefactors, as Cooley and Hemelrijk have suggested 101 . Put another way, even if the ius liberorum did not make a dramatic difference to a woman's ability to manage her own property per se (that is, to buy and sell, invest, or engage in business on her own initiative and in her own name), it might have given her freer choice in what she did with it: in what she invested in, sold, or gave away -and to whom.
One point worth emphasising, therefore, in thinking about the impact of the Augustan laws, is that we cannot generalise the position of women in tutela, even those free-born 102 . A woman like Terentia, with a freedman or some other biddable person as her tutor, probably would have gained little in practice from the ius liberorum. By contrast, for a woman in agnatic tutela, the ius liberorum could have made the difference between leaving her property to her children (or providing a bathhouse for her city, if she was so inclined) 103 , rather than seeing the whole estate pass to a greedy brother or uncle -and agnatic tutela was probably considerably more common at the time of the Augustan laws than some scholars have suggested. For freedwomen, the potential benefits were still greater, since they were more likely to be subject to tutela legitima, but we also need to distinguish between freedwomen in the tutela of their male patrons and the freedwoman of a woman with her own choice of tutor appointed under the lex Atilia. Thus, how burdensome tutela was in practice, and how much difference the ius liberorum would have made, was, to an extent, the luck of the draw, and certainly nothing to do with a woman's desire or ability to administer her own property 104 .
The Augustan laws created further inequalities, most obviously between those who produced the three (or four) children necessary for the ius liberorum and those who did not -a matter that was to a large extent outside a woman's control, especially if "three children" meant three living children. 105 . Unfortunately, it is not clear what the original 101 -Cooley 2013, 24-9;Hemelrijk 2015, 22-3. Cooley emphasises that the Augustan legislation was only one factor and part of a continuing process of change.
102 -Of course, Roman property rights were in many respects unequal. A more fundamental distinction than that between the different species of tutela, or even between male and female, was that between persons sui iuris and those still in potestate, who legally could not own any property in their own right. Most profound of all was the distinction between free and slave.
103 -Under the empire, women sometimes took care to record that they had donated land with either the consent of a tutor (as did Julia Monime, CIL 6.10231) or the benefit of the ius liberorum (e.g. CIL 6.10247, a donation by Statia Irene). For women as civic benefactresses, see Hemelrijk 2015, ch. 3. 104 -Conversely, some women seem to have chosen to act with a tutor or other male adviser even when not legally required to do so (including women in possession of the ius liberorum): see e.g. Taubenschlag 1955, 176-7;Sijpesteijn 1965, 176 on imperial evidence from Egypt. Cf. Tellegen-Couperus 2006, 434. 105 -On factors affecting fertility and implications for the ius liberorum, see Parkin 1992, law required. The problem is complicated by the fact that the Augustan laws established an array of different privileges requiring various numbers of living and/or deceased children 106 . By the third century CE, it seems that three (or four) live births were sufficient for women to claim the ius liberorum 107 . However, the original rule may well have been more demanding, as for other privileges available under the laws 108 , which also suggest a general preference for live children 109 . For instance, we know that for men, for various purposes (such as exemption from providing operae for a patron, or from acting as tutor), what counted was the number of surviving children in potestas 110 ; also, where a freedwoman gained the ius liberorum and made a will, the lex Papia granted her patron a share in her estate proportional to the number of her children who survived her 111 . Further, we know that the eligibility rules were subsequently relaxed, to some extent, through juristic interpretation and imperial enactment -for instance, by allowing men to count children who had died in war 112 , and women to count illegitimate children, contrary to the original rationale of the Augustan laws 113 . Women did not have potestas, and so the rules attested for men cannot have applied to them, but the original ius liberorum could have involved something like the "sliding scale" that gave freedom of inheritance between husband and wife: one child surviving past puberty, two to the age of three, or three to their naming-day 114 . The difference between live births and surviving children (even if only to a certain age) is significant in considering the impact of the Augustan laws, due to very high infant mortality in the ancient world. If the rule was three live births, most women would have qualified for the ius liberorum, as Kelly argues 115 . If three surviving children were needed, many would have struggled 116 . At any rate, the bar was set much higher for freedwomen, especially freedwomen in the tutela of their male patrons (or patrons' sons), who required four children to obtain the ius liberorum 117 . Moreover, a freedwoman could probably only count children born after manumission 118 , and under another Augustan law, the lex Aelia Sentia, slaves normally could not be manumitted before age 30 (Gaius 1.18). Thus, freedwomen required three or four children born after the age of 30, as against three for freeborn women, starting potentially as early as age 12. As Dixon notes, this would have made the ius liberorum virtually unattainable for the class of women most likely to benefit from it 119 .
Yet, Armani has observed that former slaves -male and female -were more likely than freeborn persons to record possession of the privilege 120 . She suggests that this may be due to differences in epigraphic habit, and the fact that freeborn persons were more likely to have other and better 114 -Tit. Ulp. 16.1. The designation of children is filium filiamve (rather than liberi) and thus equally applicable to men and women. The rule as described in this passage may not reflect the original statute either, but cf. similar formulations in Tit. Ulp. 15.2 on decuma and in the lex Malacitana, ch. 56 (the lex Malacitana incorporates elements of the Augustan marriage laws: González and Crawford 1986, 210).
115 -Kelly 2017, 114-16. He calculates that 30% of women aged over 25 in Egyptian census returns had produced three or more children, and that this figure is likely to be much too low, as 50 women are attested in Egypt with the ius liberorum between 236 and 310 CE, against only 12 with a guardian.
116 - Parkin (1992, 119) calculates that three living children required five or six births, a level of fertility that the Roman elite was evidently not achieving. The birth rate in imperial Egypt appears to have been higher (see Kelly 2017, 114).
117 -Tit. Ulp. 29.3 states that the ius trium liberorum was introduced by the lex Julia, the ius quattuor liberorum for freedwomen by the lex Papia Poppaea (cf. Jörs 1882, 26;Crawford 1996, 803). It is not clear whether the later law extended to freedwomen a privilege previously unavailable to them at all, or raised the bar by requiring four rather than three children. Eck (2019, 86 n. 19) argues that the lex Papia was in general more demanding than the lex Julia. Wallace-Hadrill (1981, 61) sees in the law an intention to diminish the status of the freedman parvenu. Perry (2014, 86-8), however, argues that the disadvantaging of freedwomen was a consequence of legislative concern with protecting the interests of patrons, rather than antipathy to freedwomen.
118 -See e.g. Gardner 1986, 20. Cf. Treggiari 1969, 214 on the smallness of freedmen's families and Dig. 38.17.2 for the rule that slave-born children did not count under the senatus consultum Tertullianum, except in special circumstances.
119 -Dixon 1988, 89. Cf. Parkin 1992 n. 128 ("virtually impossible"). 120 - Armani 2018, 16. honours to record in their inscriptions. Another reason may be that the ius liberorum was significantly harder for freedwomen (and men) to obtain, and at the same time more valuable -making it indeed something to shout about. Perhaps some managed it: the rules laid down in the lex Papia concerning succession to freedwomen's estates contemplate women producing four (freeborn) children (Gaius 3.44). It seems possible, however, that the privilege was most often obtained as a grant from the emperor, rather than by childbearing, as was the case for Cornelia Zosima, the only woman in Armani's list of known holders of the ius quattuor liberorum 121 .
The ius liberorum was innovative, but not revolutionary, and uneven in its effects: helpful to women who obtained it, but not life-changing, except for those who were thus able to escape an overbearing tutor, and probably largely unattainable for freedwomen. It is best regarded as a step in the evolution of Roman women's property rights, which had already seen the rise of sine manu marriage and the ability of women to make wills, and would see the further weakening of tutela for freeborn women, before the eventual disappearance of tutela mulierum altogether 122 . The independence the Augustan laws accorded to women was anchored in the independence many women already enjoyed in the republic, and, perhaps, the respect and responsibility associated with the Roman matrona 123 . The concept that women might henceforth operate without a tutor at all had antecedents in the position of the Vestal Virgins and the number of transactions a woman could undertake in any case without a tutor's auctoritas.
Although there is no direct evidence, it seems likely the Augustan laws anchored further (and rapid) developments in the institution of tutela 124 . The freeing of significant numbers of women from tutela must have helped to cast doubt on the value of the institution. At some stage, praetors began to compel tutors, other than tutores legitimi, to give their auc-121 -CIL 6.1877 (= Armani 2018, no. 24): Corneliae Zosimae... habenti ius quattuor liberorum beneficio Caesaris. It is usually impossible to tell from epigraphic evidence whether a holder of the ius liberorum obtained it by childbearing or by imperial grant.
122 -Tutela of adult women became obsolete before Justinian and references to it were systematically omitted in the Digest.
123 -Cf. Dixon 1988, 89, who suggests that the ius liberorum was essentially an extension of traditional republican regard for motherhood.
124 -We do hear of other legal changes consequent on the ius liberorum: for instance, both Augustus and Claudius passed edicts forbidding women from interceding (taking on debts) on behalf of their husbands, and the senatus consultum Velleianum, of Claudian or Neronian date, established a general rule (Dig. 16.1.2.pr-1; cf. Gardner 1986, 75-6). Hausmaninger (1964, 23-4) suggests that juristic developments concerning bona fides and usucapio were prompted by the new possibility of women without tutors. toritas 125 . This gave women a means (if perhaps an inconvenient one) 126 of getting their way in the face of a tutor's objections. Auctoritas could be compelled even for making a will, which was probably the single most significant legal act most women would undertake 127 . There is no evidence to date this development, which probably occurred gradually, but it would fit well between the Augustan laws and the abolition of agnatic tutela by Claudius 128 . This was the next major development in the history of tutela mulierum 129 . It meant that the only women still subject to tutela legitima were freedwomen in the tutela of their patrons and emancipated daughters with living fathers. As the latter were probably rare, Claudius' reform, in combination with the ability to compel tutors, reduced the significance of tutela for freeborn women to the point where Gaius could describe it as an antiquated and largely empty formality 130 . We do not know how Claudius introduced the measure but, in view of Claudius' fondness for historical precedent and Augustus' exemplary status, it would not be surprising if he invoked Augustus' earlier innovation in freeing many women from tutela altogether 131 .
Freedwomen, however, were left behind, still subject to old republican rules designed to protect the interests of the tutor legitimus (and new inheritance rules benefiting patrons and their descendants) 132 , with little chance of gaining the ius liberorum through childbearing. It seems Claudius was aware of their unfavourable situation: when there was a shortage of grain and he wished to encourage investment in ship-building, he offered women who built merchant ships the right of four children 133 . The specification of four rather than three children indicates that the reward was targeted, at least partly, at freedwomen 134 . In general, how-ever, freedwomen remained disadvantaged, in favour of the special privilege of patrons in Roman law 135 .
Further, although the Augustan laws helped to weaken the practical significance of tutela mulierum for many women, they provided new grounds for its maintenance as an institution. As Schulz put it, tutela was henceforth "anchored" in the Augustan laws and the ius liberorum 136 . Indeed, as far as freeborn women were concerned, and after the abolition of agnatic tutela, we might ask how far tutela mulierum was retained as a formality for the sake of granting exemption from it 137 . At any rate, the ius liberorum had a symbolic value sometimes independent of its practical effect. As noted earlier, the ius liberorum could be a source of pride, and the grant of the ius by the senate or emperor was one means of recognising and rewarding women, to whom many other forms of Roman honour were not available 138 . Moreover, by badging a set of legal privileges (including freedom from tutela) as the "right of children", even where the recipient did not have any children at all 139 , the ius liberorum served to keep the ideal of motherhood in the public eye 140 . But this symbolism should not obscure the practical advantages that the ius liberorum could bring, especially to women in agnatic tutela -perhaps still common when the Augustan laws were introduced -and to those freedwomen lucky enough to obtain it.