Published online by Cambridge University Press: 09 August 2013
I would like to thank J. Russell and B. Ward-Perkins for their comments and suggestions on earlier drafts of this paper. My gratitude also goes to E. Thomas and C. Witschel who communicated to me privately on the arguments presented here; though we remain in disagreement on various issues, their comments helped sharpen my focus. All remaining errors are my own. Abbreviations for inscriptions and epigraphic sources follow F. Berard et al. (eds), Guide de l'Epigraphiste (second edition) (Paris, 1989), 16–17; for literary sources the conventions of the Oxford Classical Dictionary are followed.
1 Thomas, E. and Witschel, C., ‘Constructing reconstruction: claim and reality of Roman rebuilding inscriptions from the Latin West’, Papers of the British School at Rome 60 (1992), 135–77CrossRefGoogle Scholar (hereafter, Thomas and Witschel). All dates are AD unless otherwise stated.
2 Thomas and Witschel, 137 (‘It will … recorded’), 139 (‘Particular … work done’), 148 (‘Except where … building’); see also below, n. 40. While symbolism and reality are by no means mutually exclusive, throughout their paper Thomas and Witschel consistently contrasted ‘notional’ or ‘symbolic’ claims with physical reality, usually affirming the former while disputing their applicability to the latter, cf. 136–7 (introduction), 140–9 (claims of destruction) and 152–7, 161–4 (various individual buildings).
3 Thomas and Witschel, 159 (cited below, n.41).
4 Thomas and Witschel, 143–7 (vetustas) and 164–75 (restituere).
5 IRT 103a (Sabratha; AD 378): ‘[Fl. Vivi Benedicti v.]p. totius integritalis moderationis iustitiae provisionis fidei benignitatis fortitudinis ac beneficentiae viro Fl. Vivio Benedicto v. p. praesidi prov. Tripol. inter cetera beneficia sua quibus omnem provinciam conpendiis remediis et virtutibus fovit sublevabit (sic) erexit etiam ob ea quae sibi specialiter conlata sunt …’.
6 According to the data supplied by Thomas and Witschel (appendix 1b (p. 175)) some 68 per cent of Latin rebuilding texts date to the period 192 to 500. The largest single percentage (34 per cent) comes from the fourth century, a time when rhetorical exaggeration was well entrenched in inscriptional vocabulary.
7 For a recent assessment of this phenomenon with ample reference to the previous literature, cf. Duncan-Jones, R.P., Structure and Scale in the Roman Economy (Cambridge, 1990), 79–104, esp. pp. 79–92.CrossRefGoogle Scholar
8 Thomas and Witschel (168 nn. 208, 209) highlighted this point when they pointed to pseudo-epigraphical language used for effect by historians (Tac, . Ann. 15. 18.2Google Scholar) or satirists (Petron, . Sat. 135. 3Google Scholar; Apul, . Met. 1. 14Google Scholar).
9 So, for instance, the appearance of vetustas in Republican and early Imperial inscriptions (cf. Thomas and Witschel, 148 n. 75) suggests that its essentially formulaic nature had been a habitual feature of the language of reconstruction by the time most rebuilding texts were carved (see above, n. 6).
10 Thomas and Witschel, appendix 1d (p. 176). If neglect, time and neglect and ruin are added, the total reaches 73 per cent.
11 Thomas and Witschel, 144.
12 For instance, many of the public buildings in Pompeii and Herculaneum are faced with plaster, as were once the impressive Baths of Agrippa in Rome (Pliny, , HN 36. 189Google Scholar); a specific verb (calicare) described the process, cf. Festus 41L (47M); CIL 22.1529 = 10.5807 = ILS 5348 = ILLRP (Degrassi) 528 (Aletrium; late second century BC); on the pitfalls of plaster facing, cf. Vitr., De Arch. 2. 8.20.Google Scholar For a recent modern treatment, cf. Adam, J.-P., Roman Building: Materials and Techniques (Bloomington (IN), 1994), 224–7.Google Scholar
13 Cf. TLL, 3.1573.17–43 s.v. where collabor can denote a descent into a degenerate state (notably associated with old age, cf. Cels. 7. 26.1), not necessarily a sudden collapse.
14 As claimed by Thomas and Witschel, 145 esp. n. 57, where they took as an illustrative example the discrepancy between two records of the same bath restoration at Telesia — one mentioning an earthquake (AE 1972.150) the other not (CIL 9.2212 = ILS 5690; Thomas, and Witschel, cited CIL 9.3018Google Scholar = ILS 5761 here). But this is an argumentum e silentio: any number of factors could account for omissions in the individual wording of texts recording the same action. Compare the fragmentary versions of a text on opposite sides of the same slab from Thuburbo Maius commemorating the completion, within seven months, of restorations to the summer baths that had been in progress for eight years (ILAfr273; AD 361). The difference in wording between the two texts is analogous to that at Telesia: one (273b) omits lengthy descriptions of the building's poor state (that is, the reason for the restoration) found in the other (273a). Does it follow that the damage as described was ‘notional’?
15 To be fair to Thomas and Witschel, they conceded (145) that such claims may be ‘broadly accurate’, but also argued that they are ‘highly symbolic’, citing the example of the baths at Telesia (see previous note). Their assertion that ‘there are signs of a will to exploit the event of earthquake as a factor of destruction, even if it might not have been the real or only cause’ (Thomas and Witschel, 145; my italics) surely allows only two possibilities: either there was earthquake damage, and the restorer claims to have fixed that damage, or there was not, and the restorer makes a false claim.
16 Discussed by Thomas and Witschel, 145–6.
17 CIL 13.11759 = ILS 9179b (AD 241): ‘aquam Gordianam … multo tempor(e) [inter]missam …’.
18 CIL 13.11758 = ILS 9179a (AD 231). Thomas and Witschel (146) concluded that this shows the phrase multo tempore carries here a ‘purely notional value’.
19 The commanding officer at the time of the Gordianic restoration is named in the text (above, n. 17) as C. Jul(ius) Roga[tianus e]q(ues) R(omanus). Note that the water was led specifically in praetorium and in balneum, both places commanders would have an interest in making as comfortable as possible.
20 For example, baths: ILS 5693 = Eph. Epigr. 8.45 6 (balneum … lon[gi temporis] vetustate corruptum); CIL 10.222 = ILS 586 (balnea … po[st longam] seriem ann[orum resti]tuit); AE 1984.151Google Scholar ([thermas long]i temporis deformatas).
21 Cf. n. 42 below for the phrase ‘the baths … for so many years hence deformed by the decay of ruins’ (‘thermas … tot re]tro annis ruinarum labe deformes’). Some restoration texts (especially of baths) highlight the positive results of reconstruction when they stress the return of the building to the use of the community, for example, CIL 14.5387 (lava[tu]m … ad usum pop[uli]); CIL 5.7250 = ILS 5701 (thermas … usui Segusinae reddidit civit[atis]); Notizie degli Scavi di Antichità (1928), 283 (balinea … refecta in usu mu[nicipum or municipii]).Google Scholar
22 Thomas and Witschel, 159–64. A fourth-century inscription that commemorates the construction of a tribunal at Saepinum a fundamentis (CIL 9.2448), was declared ‘doubly misleading’ (164 n. 177) because the lower section is clearly Augustan and the work done was a reconstruction, not an initial construction (163–4). Here, Thomas and Witschel clearly interpreted a fundamentis quite literally.
23 As argued convincingly by Thomas and Witschel, 149–64.
24 For example, Thomas and Witschel (160 n. 149) offered ‘an almost nonsensical example’ to discredit the term a fundamentis when they cited a text that mentions an ‘ocean begun from the foundations’ (AE 1975.873: ‘oceanum a fundamentis coeptum.’). They interpreted oceanus literally to declare it ‘almost nonsensical’, but also postulated a symbolic significance for the term. The latter is probably closer to the truth. The word is best interpreted as referring to a named part of the building, probably a bath-house (cf. , S.H.A.Alex. Sev. 25.5Google Scholar; AE 1989.743); it may have been a room or a pool, decorated with marine scenes of a sort familiar to archaeology, cf., for example, the Baths of Neptune at Ostia (Yegül, F., Baths and Bathing in Classical Antiquity (Cambridge, MA 1992), 68 and esp. fig. 75 on p. 71).Google Scholar If so, oceanus could indeed be a structure ‘begun from the foundations’. For rooms in baths named for decorative features, cf. those at Gader, Hamat — Green, J. and Tsafrir, Y., ‘Greek inscriptions from Hamat Gader’, Israel Exploration Journal 32 (1982), 77–96.Google Scholar
25 Cf., for example, Snodgrass, A.M., ‘Archaeology’, in Crawford, M. (ed.), Sources for Ancient History (Cambridge, 1983), 137–84, esp. pp. 143–6Google Scholar; Finley, M.I., Ancient History: Evidence and Models (New York, 1985), esp. pp. 27–46.Google Scholar
26 Thomas and Witschel, 144 and 152–327
27 CIL 13.659 2 = ILS 9184; balineu[m] vetustate conlapsum expl(oratores) … et al. … de suo restituer(unt). The text is firmly dated by the consular formula Lupo et Maximo cos (sic).
28 Baatz, D., ‘Das Badesgebäude des Limeskastells Walldürn (Odenswaldkreis)’, SaalburgJahrbuch 35 (1978), 61–95.Google Scholar
29 Note that the excavator identified a restoration and alteration of the new bath due to weaknesses in the suspensura in the caldarium, cf. Baatz, , ‘Badesgebäude’ (above, n. 28), 87–9.Google Scholar This repair cannot be dated precisely and should be considered a candidate for the vetustate conlapsum commemorated in the inscription.
30 Coin finds suggest that the old bath was in use until some time after 200, but exactly when it was burned and rebuilt is not datable by similar means (it could conceivably have burned down after only a few months of use), cf. Baatz, , ‘Badesgebäude’ (above, n. 28), 92.Google Scholar Indeed, Baatz (on pages 62, 66, 74, 91–2) dated the fire and major reconstruction chiefly on the testimony of the inscription, a practice which proved convenient to them here but which Thomas and Witschel roundly condemned elsewhere in their paper (139): ‘Archaeological evidence is often unscientifically assessed … It is almost never used independently of the rebuilding inscription, and this results in arguments based on obvious circularity’.
31 If baths required constant maintenance and major work every 20 to 30 years, as Thomas and Witschel postulated (159), then it would be remarkable for only one inscription to have been set up at Walldürn during the almost 100-year period between the baths' initial construction and the dated text; compare the numerous commemorations from the (admittedly grander) Hadrianic Baths at Lepcis Magna (IRT 263, 396, 794b) or the Summer Baths at Madauros (ILAlg 1.2101, 2102, 2108); both were discussed by Thomas and Witschel (159 and 162). Given this, it is reasonable to suggest that there may have been other texts, now lost, associated with the Walldürn bath-house. The association of the inscription with the major act of reconstruction at Walldürn is therefore not nearly as certain as Thomas and Witschel presented it. For other instances where a text cannot be related to archaeologically attested reconstruction with absolute certainty, cf., for example, Thomas and Witschel, 147–8 (temple of Liber Pater) and 153–4 (fort at Vindolanda).
32 Cf., for example, Thomas and Witschel, 153 (Jagthausen baths), 153–4 (Vindolanda fort), 154 (Risingham fort), 163 (the praetorium at Cologne).
33 Thomas and Witschel, 139–40.
34 CIL 6.896 = 31196 = ILS 129.
35 de Fine Licht, K., The Rotunda in Rome. A Study of Hadrian's Pantheon (Copenhagen, 1968), 237–45.Google Scholar
36 Thomas and Witschel, 156; see also 166–7. The wording of adornment phrases varies but can be found in many texts Thomas and Witschel referred to in their paper, usually without comment: e.g. CIL 5.7250 = ILS 5701 (ornavil); CIL 8.20836 = ILS 638 (ad pristinum statum); CIL 10.6656 = ILS 5702 (in meliorem civitatis effigiem); AE 1937.119 (at pulcridinem [sic]); AE 1985.876b (exornatam); ILAfr 506 (excultas); ILAfr 275 (ad melio[rem] cul[tum]); IRT 396 (marmoribus et co[l]umnis exornavit); IRT 543 (in meliorem faciem).
37 Cf., for example, Pliny's bequest to Comum (CIL 5.5262 = ILS 2927) which included HS300,000 for the decoration of the baths there — the equivalent of the entire cost of constructing a bath (no doubt on a smaller scale) as estimated to Fronto (Gell, . NA 19.10.1–4Google Scholar). The same figure was spent at Epamanduodurum in Germany on decorating the baths with marble (CIL 13.5416), and at Corfinum a bath that had been built and roofed by a deceased benefactor required a further HS252,000 to complete, presumably mostly for decoration (CIL 9.3152 = ILS 5676). On the great expense of sculpture, cf. Marvin, M., ‘Freestanding sculptures from the Baths of Caracalla’, American Journal of Archaeology 87 (1983), 347–84, esp. pp. 380–1.CrossRefGoogle Scholar
38 Thomas and Witschel conceded these points, but paid little attention to them in the main body of their argument, cf. 139, 155, 171 (decoration and the ancient visitor) and 139 (survival).
39 De Fine Licht's discussion of the surviving ancient decoration of the building is marked by uncertainty over what is ancient and what is not, cf. Licht, de Fine, Rotunda in Rome (above, n. 35), 102–26, 142–6.Google Scholar That the marble veneer of the upper level of the interior (torn out in 1747) can be shown to have been Hadrianic hardly discredits Severus's and Caracalla's claim (as Thomas and Witschel contended (135–6)): the exact nature of the ancient decoration of the building throughout the Imperial period cannot now be reconstructed, so there is no way of knowing how many decorative elements were added in ancient times and subsequently removed. Just two examples: the floor must have undergone several renovations over the years (most recently in 1873), so that even if the pattern is original, the materials are not; and the tabernacles in the lower zone of the interior show signs of numerous alterations — on one occasion in 1597 their columns, marble facing and metal elements were stolen and sold by the men entrusted with the care of the building (cf. Licht, de Fine, Rotunda in Rome (above, n. 35), 101 and 111 respectively).Google Scholar
40 Despite the occasional disclaimer that some texts may reflect real damage (for example, 144, 145, 148), the main thrust of Thomas and Witschel's argument is in the opposite direction, thus (159): ‘[the] purpose [of rebuilding inscriptions] was more to record symbolic reconstruction than to provide accurate statements to be checked against the actual effects of the operations’ (my italics).
41 On which, cf. Thomas and Witschel, 159: ‘There is no reason to believe that inscriptions become any more denotative when they are filled with minor details’.
42 ILAlg 1.2101 (AD 364): ‘pro tanta securi[tate temporum] dd(ominorum) nn(ostrorum) Valentiniani [et Valentis perpetuo]rum Aug[g(ustorum) therm]as aestivas, olim splen[did(issimae)] coloni[ae nostrae? orn]a[mentum? sed? tot re]tro annis ruinarum labe deformes pa[rietibusque omni?]um soliorum ita corruptis ut gravibus damnis adficerent, [nun?]c omni idonitate constructas et cultu splendido decoratas, sed et patinas ampliato aeris pondere omni idonitate firmissas … Cec. Pontilius Paulinus, ff(lamen) p(er)p(etuus), p(atronus) c(oloniae), curat[o]r rei p(ublicae) pecunia publica perfecit…’.
43 Thomas and Witschel, 159: ‘The language of ruins cannot, then, have been literally appropriate to the whole building, if, indeed, it was for any of it’.
44 The appearance of the curious word patina in this text surely points to the essentially accurate description of the work done. It is a rare word in any context, but it has here the feel of a technical term, designed perhaps to impress the reader with the thoroughness of the benefactor's efforts. Today, patinae (possibly associated with the boiler apparatus, cf. Palladius 1. 39.3) cannot be securely identified in this or any other remains of baths.
45 Thomas and Witschel, 164–75.
46 Romans did not like people who claimed to be what they were not, cf., for example, Martial's relentless savaging of Ligurinus, the aspirant poet (3. 44, 45, 50; compare Petron. Sat. 92), or of the nouveaux riches such as Aper (Mart. 12. 70) or Zoilus (2. 16, 19, 42, 81, 3. 29, 82, 4. 77, 5. 79, 6. 91, 11. 12, 30, 37, 54, 85, 92, 12. 54). Lucian also lampooned false claims to status, noticeably in Alexander of Abonuteichus, and Pliny showed contempt for false public display, for example, Ep. 4. 2, 4. 7, 6. 5. 5–7.
47 Amm. Marc. 27. 3.7; for Lampadius, cf. PLRE 1 Volusianus 5 (pp. 978–80). This anecdote supports Thomas and Witschel's portrayal of misleading inscriptions; but it also shows that such mendacity was not appreciated.
48 The seminal work on euergetism is still Veyne, P., Le pain et le cirque. Sociologie historique d'un pluralisme politique (Paris, 1976).CrossRefGoogle Scholar (For an abridged English translation by Pearce, B., see Bread and Circuses: Historical Sociology and Political Pluralism (Harmondsworth, 1990).Google Scholar) Thomas and Witschel held a different view of the function of inscriptions: they were aimed at posterity so that ‘the vagaries of the building's previous history and the critical importance of its reconstruction are not contemporary events known to the general reader, but historical facts necessary for public architectural education’ (167–8). I believe that this interpretation takes too narrow a focus. Certainly, posterity appears to have been an important factor in motivating the erection of inscriptions, but their immediate social value to benefactors cannot be overlooked. (Of course, which of the two considerations — distant posterity or immediate present — took precedence in benefactors’ perceptions remains open to question.)
49 This applies also to inscriptions put up by benefactors on their own buildings (rather than, say, those carved on a statue-base by the local decurions): here, the text represents the benefactor staking a claim to the prestige that was his or her due. Often, of course, benefactions could be commemorated by both the benefactor (on the building) and the beneficiaries (on statue-bases and the like), cf. the example of the winter baths at Ocricoli, the restoration of which was commemorated both in a building inscription (CIL 11.4095 = ILS 5696) and on two statue-bases dedicated to the restorers (CIL 11.4096 and 11.4097 = ILS 5697). Cf. Thomas and Witschel, 167.
50 For example, CIL 3.1805 = ILS 5695 (meal; Narona, AD 280); CIL 2.1956 = ILS 5512 = ILER 2054 (meal and games; Cartima, Flavian?); CIL 10.5918 = ILS 406 (hand-out and meal; Anagnia, late second century).
51 Many examples of such fiscally informative texts are collected by Duncan-Jones, R.P., The Roman Economy: Quantitative Studies (second edition) (Cambridge, 1982)Google Scholar, nos. 1–212 (North Africa) and 439–549a (Italy). These texts often deal with initial construction, but some are concerned with restorations: for example, CIL 9.6075 = ILS 5875 (HS1, 726, 100 spent on a road vetustate amissam); ILA.fr 403 (HS200,000 spent on a building and porticoes vetustate co[nsumpt.]); CIL 11.3126 = ILS 5374 (HS100,000 spent on two roads vetustate consumptas); CIL 10.6309 (HS100.000 spent on restoration of a Temple dedicated to Tiberius); CIL 11.3123 = ILS 6587 (HS29,300 spent on restoration of a portico vetustate dilapsam); AE 1919.64 (HS14,000 spent on a road vetustate reciss(am)); CIL 5.6513 ([lost] amount spent on bath vi consumptum).
52 IGRR 3.739 = TAM 2.905. For a translation and discussion of this huge text, cf. Danker, F.W., Benefactor: Epigraphic Study of a Greco-Roman and New Testament Semantic Field (St. Louis (MO), 1982), no. 19 (pp. 104–51).Google Scholar
53 Just three examples: CIL 9.3430 = ILS 5668/9 (balineum refectum … pecun. public); CIL 10.5055 = ILS 5349 (clovacam (sic) reficiundam d(e) s(ua) p(ecunia) c(uravit)); CIL 8.7957 = ILS 5408 (templum … sua peq(unia) renovavit).
54 Cf. DeLaine, J., ‘Baths and benefactors in Roman Italy’, in Johnston, D.E. and DeLaine, J. (eds), Roman Baths and BathingGoogle Scholar (forthcoming in the Journal of Roman Archaeology Supplementary Series).
55 ILTun 1500 = ILAfr 573; cf. AE 1925.31 (AD 367–83): ‘atrium thermar[um Lic]inianarum ab antiquis c[oe]ptum … quod inperfecto opere corruptum adque ruderibus foedatum [erat …]dius Honoratinus fl. p. cur. reip. II [cu]m statua signoq. felicissimi Fl. Gr[atian]i CCCRATU opere perfecit. The letters CCCRATU were incomprehensible to Merlin (ILTun), but probably read something like [a]ccuratu (sic) opere, in contrast to the imperfectum opus of the antiqui. The original construction dated to the reign of Gallienus, AD 264 (CIL 8.26559 = 15521 and 15246a, b = 10620 = ILTun 1416); for a recent description of the physical remains, cf. Yegül, , Baths and Bathing (above, n. 24), 206–12.Google Scholar