What’s in a number? According to Elle Hunt’s recent feature in the Guardian, ‘it is dull to reduce human experience to numbers’. More specifically, Hunt ridicules the fact that ‘Spotify knows how many hours I spent listening to Taylor Swift’—because ‘only I know why’—concluding that ‘(t)he insights resulting from this data are close to zero’. Central to Hunt’s derision of the criticised number-crunching is the contention that context matters for any meaningful interpretation of quantitative data: without ‘knowing why’, the numbers remain meaningless.
In their study of the ages recorded in Latin-inscribed epitaphs from (mostly) the Roman imperial period, Ray Laurence and Francesco Trifilò aim to give meaning to a pool of numbers that is both vast and simultaneously unusual in ancient funerary epigraphy, including in Latin epigraphy. Discussing the age patterns produced by over 20,000 inscriptions, Laurence and Trifilò seek to contextualise age-at-death data especially from a cultural perspective: ‘(w)hat is needed is an overall understanding of the way we might look at age as recorded in all epitaphs across the Latin West’ (2). To this end, they compare ‘patterns of the practice of age-commemoration between cities and between regions to determine how age was used’ (7), thus to map out Mediterranean Timescapes around the western half of the corrupting sea.
The numbers at their disposal stem from what are mostly simple epitaphs that list (typically) the names of the deceased, give (sometimes) information about the commemorators and (always—given the remit of their study) the (supposed) age-at-death of the deceased. An example for the kinds of epitaphs under scrutiny is the tombstone of Iulia Senica from Great Bulmore in Wales, just east of Caerleon (RIB 374):
D(is) M(anibus)
Iulie Senice
vics(it) an(n)os LX
(To the divine manes of Iulia Senica; she lived 60 years.)
Like this epitaph, the inscriptions studied in Mediterranean Timescapes present what Laurence and Trifilò call ‘the age-related formula vixit annis’ (41), including variants (such as in Iulia’s epitaph) and abbreviations (e.g., ‘vixit A or vixit An/n or just VA’: 9). Although focused on the Latin West of the Roman Empire, rich in funerary epigraphy, Dalmatia, Rome, Corsica and (going by the maps) Britannia are excluded (4-5)—and therefore in fact Iulia and roughly another 20,000 individuals commemorated with age-at-death in these places:[1] the remainder, Laurence and Trifilò contend, ‘provide sufficient breadth for the task of the project’ (3).
From the start, the emphasis on the cultural dimension of funerary commemoration is mobilised to distance the study from demographic analysis: ‘demography and the commemoration of the age of death are two quite separate things’ (24). Consequently, thirteen chapters across three sections explore diverse cultural aspects, such as perceptions of old age, the broader societal contexts of age rounding, the gender bias behind the greater epigraphic visibility of males, the motivations for commemoration by kin, the role of the military in inscribing numbers—the last seen as the driver for the spread of age commemoration (228). Three key themes serve to illustrate the manner of exploration and argumentation: geography, status and, naturally, age.
First, geography. The discussion reinforces repeatedly the differences in epigraphic behaviour between Africa and Europe—concerning commemorative language (e.g., 56-65) and age recording (e.g., 62-65, 94-7; 99-103). The African provinces preferentially employed the abbreviation ‘VA’, with Numidia using it roughly 90% of the time, while other provinces privileged a form of ‘vixit a/an/ann’; Italy (and Sicily) present the most balanced usages of ‘the formula vixit annos/annis […] its contraction and abbreviation’ (58). Furthermore, the African provinces head the commemoration of older individuals, including people aged 100 and over, while especially Italy’s commemoration habits appear concentrated on children and young(er) adults, roughly up to age 30. Unlike with Italy, the African life trajectories do therefore not correspond ‘to modern expectations of pre-modern mortality’ (103), underscoring the noted rejection of demographic interpretation. Instead, the African patterns are understood as prizing longevity, thus ‘to institutionalize age and expectations associated with age’ (103)—without, however, offering analytical comment on why children and younger individuals were at all commemorated by age in this cultural climate, or explaining the attested concern with older ages also in Italy. Stressing the need for a less Eurocentric approach in ancient world studies (e.g., 8), this geographical distinction between Africa and Europe—which ‘had been known previously’—informs the discussion throughout ‘to reveal the local or regional variation that underlies the use of a global phenomenon’, i.e. the recording of age in epitaphs (56).
Second, status. Having selected ‘Social status (as stated in the epitaph)’ as a criterion for study (4), in effect focused on dimensions of legal status (cf. 12, n. 12), 420 inscriptions of freedpersons identified ‘with the word libertus/liberta’ (130) unsettle the neat age differentiation between Africa and Italy presented elsewhere: while Italy commemorates mainly younger freedpersons, with a notable cluster in the early 20s, both geographies demonstrate comparable commemoration patterns of freed individuals aged 60+ (137-8). The (muddled) discussion of the role of age 30, pertaining to the so-called Augustan manumission legislation, and the status of Junian Latins—individuals manumitted without Roman citizenship—highlights conversely the comparatively small number of imperial liberti (i.e., freedmen) commemorated before age 30: ‘30 would appear to be a significant break point in the data set […] perhaps due to knowledge of the lex Aelia Sentia’ (133-4). Comment is lacking here on the widespread view that manumission by the Emperor always produced enfranchised freedpersons, irrespective of age. The statement that a woman freed iusta causa for the purpose of marriage (i.e. through manumissio matrimonii causa: e.g., Gai. Inst. 1.19) was thus ‘established as free, but as a Junian Latin tied to the household with her estate returning to her patron or his heirs at her death’ (133) similarly surprises. More interesting is that the commemorated libertae/i broadly mirror the epigraphic behaviour of the Italian population at large (134-9), with Italy being confirmed more generally as a hub for age commemoration of the freed, prompting the comment that ‘slavery was much more developed in Italy with a much stronger mechanism for the freeing of slaves’ (138-9).
Thirdly, age. Discussion of multiple approaches to the study of (chronological) age, including ancient conceptualisations thereof (33-40), is followed by recurrent comment about the meaning of different ages and age stages at different places, such as the already noted appreciation of longevity in Africa. This appreciation is also seen in the emphasis on offspring-to-parent commemoration in the 60+ age groups in the African provinces, in contrast to (south-central) Italy, again: ‘longevity was a feature to be cherished in North Africa, whereas the loss of family members and the dislocation of the nuclear family were dominant features that generated the patterns found in Latium and Campania’ (119-20). Comparing age-at-death and years of service in the military, demonstrating typical recruitment in the late teens and early twenties, serves conversely to argue for ‘a societal pattern that regarded children as becoming adults in the age range of 18 through to 21’ (169). The suggestion is not contextualised through the previously discussed minimum age for legal marriage, i.e. 14, for males (and 12 for females), and the epigraphic evidence for early marriage (24-5; 94), thereby effectively implying that marriage was not a sign of adulthood.
The Herculean task attempted by Laurence and Trifilò cannot be overestimated: the seemingly simple funerary epigraphy at the core of their study is fiendishly tricky to handle, even more so when the numbers are high. Confusions over the (presentation of the) data are not unexpected therefore, from the data pool—23,227, 23,723, 26,000 (3; 110; 169)—to figure keys—e.g., husband-commemorators of males, wife-commemorators of females (207-8). But the visuals are badly chosen to present the data, with many figure keys hardly legible, and numbers and percentages regularly to be gleaned just from tiny graphs. The poor copy-editing and proof-reading makes engagement with the work additionally difficult and time-consuming. I also found the focus on Anglophone scholarship (sometimes stated: e.g., 65) troublesome vis-à-vis epigraphy’s profound international connectivity and collaborativeness. There is also a distinct interpretative superficiality: several suggestions-cum-conclusions merely describe the observed patterns, often lacking in-depth exploration and argumentation, practically evading Elle Hunt’s stress on the importance of ‘why’—and on probing deeper why—for gaining insights from numbers. Notably, without comparison with epitaphs that do not include ages at death (i.e., the majority), interpretation of this datum is surely always compromised?[2]
Mediterranean Timescapes raises several interrelated questions about epigraphic data collection, identification and management:
1 Data collection
The need to detail the method of data collection is partially met, specifying the database searched, Epigraphische Datenbank Clauss-Slaby, and the categories of interest, including date, gender, and status (esp. 3-4). Brief comment about a ‘complex semi-automated system of data collection and compilation’ (240) is, however, not further expanded. Questions that, when answered, enable the assessment of the soundness of the collected data remain open. Notably: was the full epigraphic data from each region/province read manually to identify relevant texts, or merely electronically searched (see also below)? How were decisions on ambiguous inscriptions and exclusions reached? Were the texts checked against other editions/corpora/data repositories?
2 Data identification
The chosen search term is readily explained, with ‘the formula vixit annos/annis’ (58) constituting a recognised tool to identify texts with ages at death, producing a sizeable data-set. The particular search methods require equal explanation, especially regarding the relationship between manual (human) and electronic (machine) searching. If (only) electronically searched, were ‘vix’ and ‘ann’ searched together, or separately, or were perhaps only the full words used (‘vixit’, ‘annis’, ‘annos’)? And was ‘liberta/us’ searched in all cases and numbers? Comment is also lacking on the seeming exclusion of other terms, especially ‘annorum’, but for a lonely mention (at 80) of a single, grammatically incorrect usage of ‘annorum’ in combination with ‘vixit’ (IRT 1121, but referenced merely to a modern article and omitting comment on the wider use of ‘vixit annorum’, not least in Africa: e.g., IRT 668, 990 and 1136). Usage of age-at-death-formulae across the Empire was regionally varied: Rome and Italy revelled in ‘vixit annis/os’, while the northern arc from Germania to Dalmatia (marginalised in the discussion), and the Spanish provinces, privileged ‘annorum’ (see the table below). Without the ‘Big 4’ (Africa, Rome, Italy, Spain), the usage is near equal: 48%/52% (4444 vs. 4780 texts). The focus on ‘vixit annis/os’ introduces a Rome- and Italy-centred perspective, simplifying the Romanisation debate (e.g., 209-18). Status identifiers like ‘liberta/us’ were employed similarly heterogeneously (and what about the epigraphic behaviour of libertae/i not labelled thus?). How is such bias, regional or otherwise, to be mitigated for historical interpretation if only selected search terms are employed for data identification?
3 Data management
The need for digital data storage to be updated to remain usable in the future is highlighted (240-2). The need to make inscriptional big data accessible to the research community is also acknowledged (e.g., 8; 12, n. 11; 242), but not accomplished. Even if only enabled in a proprietary format, which does not depend on costly data management, such access is the gold standard.[3] Lack of access disqualifies the data from becoming ‘a jumping-off point for others with new methods and new skills’ (242), disallowing moreover analytical judgement of the data validity and robustness for interpretation. Figures and graphs (even when suitably sized) cannot replace due data access.
Sharing information on data collection and identification remains critical for epigraphic research, while the usability for historical interpretation of inscriptional big data collected through digital tools depends on familiarity with the evidence and the analytical choices made in the collection and identification process. The idea to establish ‘an absolute data set of Latin inscriptions in the next few years’ (emphases added) through machine learning classification ‘to create a single, rather than multiple sources of truths’ (241) obscures the role of necessary, human decision-making in the process of establishing robust datasets and meaningful interpretations—decisions that will by necessity differ from scholar to scholar. Elle Hunt’s emphasis on knowledge of context in the usefulness of number-crunching applies also to epigraphy.
Notes
[1] 20,000 is the (smoothed) number derived from searching (1) ‘vix’ and ‘ann’ and (2) ‘annorum’ for these locations on EDCS (31/12/23), with overlaps and 10% potentially fragmentary age data subtracted.
[2] There is also little broader historical comparative study, or anthropological and sociological contextualisation.
[3] An exemplary case is Banducci’s Foodways, reviewed in BMCR 2022.02.07 (with links to the data repositories).