My response to Riggsby’s queries concerning my review of his book, Mosaics of knowledge, aims to bring clarity to an important subject for scholars of the Roman world. As noted in my review, my main problems with the book are its selective and often misleading treatment of sources and the omission of large amounts of documentary evidence that falsifies some of the author’s central hypotheses. I will address specific points raised by the author and comment on general issues.
Riggsby questions my observation concerning minimalistic references to properties in census declarations and the Trajanic alimenta tablets[1] as examples of what he calls “obligatory cross-referencing” on the grounds that his definition thereof is “expressly about the use of numerical coding to move from one document to another.” Such coding need not be numerical, depending on how the records in question were organized. Surviving evidence for Roman records of property ownership shows that these were ordered by personal names and topographical locations, so no numerical reference (e.g. to tabula, pagina, etc.) is to be expected.[2] The name and residence of the individual under which the property could be looked up in the relevant records is what constitutes the cross-reference. As an example: the declaration “we own a new house in which we live and register ourselves, located in the same quarter of the city in which we registered ourselves in the last census” is meaningless without information contained in other administrative records, and this is precisely how Riggsby defines obligatory cross-referencing.[3]
Riggsby’s argument that the Roman state was incapable of standardizing weights and measures (83-129) was only briefly mentioned in my review due to issues of space. My criticism concerns Riggsby’s frequently inaccurate or misleading use of evidence in building up this argument. I must limit myself to only a few examples. In the two articles cited by Riggsby as evidence for “at least five” standards of modii (92), Duncan Jones concludes that literary, metrological and documentary sources offer mutually consistent information about two distinct modius measures that were widely used in the Roman empire —the italic modius and modius castrensis—the second of which existed in a flat and heaped form attested in papyri.[4] Riggsby’s claim that “there are fewer terms than standards, and in any case the words are used only sporadically. Hence, any given case is ambiguous” (92) misrepresents these findings. Similarly, Riggsby’s account of the famous inscription about grain shipments from Arles (CIL III 14165,8 = ILS 6987) misconstrues its contents, stating that the navicularii were “required to bring with them marked iron measures and appropriate personnel to reproduce the measurements on arrival in port” (110). In fact, it was the Roman procuratorial administration that was tasked with engraving iron rods (regulae) and sending its officials to supervise and accompany the shipment (imprimi charactere regulas ferreas et adplicari prosecutores ex officio tuo iubeas qui in urbe pondus quod susceperint tradant). Riggsby’s description of this as “bring your own and show your work,” which he treats as evidence for the lack of standardized measures, is accordingly incorrect. The exact function of the iron regulae has been debated since Mommsen, and the discussion of Carla Corti (cited by Riggsby) takes up Mommsen’s hypothesis that the rods functioned as a seal on the cargo with an inscription of its exact quantity.[5] It is equally plausible for the engraved rods supplied by the procurator’s office to have dictated an official standard for the measurement of the cargo. The inscription offers no indication that the standards used by procuratorial officials in charge of the annona at Arles were different from those employed at Rome.
Riggsby’s analysis of bronze weights calibrated on official Roman standards (exactum ad Castoris, 100-103) reaches the staggering conclusion that discrepancies of 10% or even 20% were “common” for official weights (101).[6] In fact, of the sample of 71 weights, 47 (two thirds of the sample) deviate from the standard by less than 5%, of which 23 (one third of the sample) deviate less than 2%. The rest of the sample contains several weights without the exactum label and 9 outliers that fall below the standard by more than 20% and include a number of significantly underweight specimens within otherwise well-calibrated sets of nested weights, whose deviation cannot plausibly be attributed to erratic weight standards.[7] If these are excluded from the calculation, the remaining sample of 56 weights yields an average deviation of 2.89% with a median of 2.1%. This is a remarkably high degree of precision for bronze weights subject to use-wear, corrosion and damage over the course of two millennia.[8] There is nothing in this evidence to support Riggsby’s claim that discrepancies of 10 or 20% were common for official Roman weights.
One generally wonders why Riggsby’s discussion of Roman weights focuses on bronze and lead weights, which are regarded as the least reliable. As noted by the numismatist Philip Grierson in his well-known lecture on weight standards in Roman coinage: “lead and iron weights are usually so badly corroded as to be useless; bronze weights are often corroded or encrusted, and may have lost the silver inlay of their inscriptions; only glass and stone weights can be regarded as really reliable.”[9] But stone weights are only cursorily mentioned (103) and glass weights—which were made in tiny denominations to weigh individual coins—not at all. It has always been clear to numismatists that the consistent weight standards of Roman silver and gold issues—which are routinely used to calculate the theoretical value of the Roman pound—imply the availability of precisely-calibrated weights in Roman mints throughout the empire.[10] But this fundamental argument for the existence of standardized weights and measures finds no mention in Riggsby’s book.
Riggsby’s use of the Roman legal sources on weights and measures is similarly problematic. The possibility mentioned in legal texts for parties to agree on specific measures (e.g. Dig. 18.1.71) does not support Riggsby’s claim that no standard measures existed (“parties can clearly opt for any standard, specific or general, that they wish, so no weights and measures can be inherently iniquum,” 111). It seems obvious that sanctions against individuals who employ weights and measures that are iniqua, maiora, minora or falsa (Dig. 4.3.18.3; 11.6.5.2; 19.2.13.8; 47.2.52.22; 47.11.6.2; 48.10.32.1; 48.19.37) are predicated on the possibility of verifying whether weights and measures correspond to their stated value, which implies the existence of official standards of measurement. That iniquae measures were destroyed by officials (e.g. Dig. 19.2.13.8; Pers. 1.129-130) or, in a striking example from Roman Philippi, melted down and turned into a public monument to aequitas Augusta, underscores the perceived importance of such standards, as does archaeological evidence for weights being recalibrated.[11] Riggsby downplays the significance of “publicly verified” (publice probatae) measures for grain, wine and other commodities in Dig. 48.10.32.1 by claiming that these were limited to the annona (108). But the text of Modestinus excludes this interpretation: the users of the mensurae publice probatae were clearly private individuals, whereby any buyer or seller guilty of tampering with them could be penalized in duplum under civil law and banished to an island if the matter were pursued criminaliter.[12] The very possibility of weights and measures being falsa(e) and prosecuted as forgery (47.11.6.2; 48.10.32.1; 48.19.37) constitutes clear evidence for the existence of official standards. One can hardly be banished to an island for falsifying something that does not exist.
Riggsby objects that I misconstrue his statement about indexing marks in Roman records being “more about authority than recall” (203) as signifying his lack of awareness that such marks were tied to robust practices of looking up documents in archival records and producing authenticated copies. As such practices are nowhere discussed in the book (nor indeed in the 1991 Hopkins paper that Riggsby cites) and run counter to Riggsby’s negative assessment of Roman archival capabilites (see e.g. “not organized well for later use” in the same footnote, 203)[13] it was logical to infer that he was not aware of this material. Are we then to understand that evidence of highly functional indexing systems and standards of retrievability in the Roman imperial context was omitted intentionally?
As regards the question of archival searching, about which Riggsby expresses doubts, in addition to efficient systems of indexing and retrieval being an achievement of great significance in the Roman context, an acid test of archival capabilities is the capacity of administrators to conduct general searches of archival materials. This has often been doubted for the Roman empire, and yet there is clear evidence for it. I limit myself to three examples. In the year 169, an official in the provincial capital of Alexandria was asked to search tax records from the last twenty years—which is explicitly described as a straightforward operation from records that were “at hand”—to check whether the association of fullers and dyers had ever paid higher taxes than were indicated in the provincial tax code (gnomon) and confirmed that no alternative sums were recorded.[14] For the Hadrianic antiquarian Phlegon of Tralles to compile a list of centenarians residing in Italy and several Roman provinces, which Phlegon says he obtained from census records, would most probably have necessitated a search through population lists organized by year of birth.[15] In the year 128 a Hadrianic prefect of Egypt was puzzled by a local Egyptian usage and ordered a search of his administrative records to find out when this issue was first legislated on; his staff searched the records and found an edict of a Flavian prefect on this subject from forty years earlier.[16]
Riggsby responds to my objection to his large-scale omission of documentary evidence (mainly but not exclusively papyrological) relevant to his study by arguing that (a) some of the evidence that I say is missing is in fact included, while (b) other evidence of which he is aware has been intentionally excluded as irrelevant or redundant.
As examples of evidence supposedly included, Riggsby cites his discussion of tomoi synkollesimoi and tabulae/paginae on pp. 29-31. These pages contain cursory descriptions that mention only a few of the numerous forms of documentation in which nested listing techniques were used (“I know of five sets of documents arranged in this way,” 29) with no mention of the census, official correspondence and judicial proceedings, notarial records, bank records, land registers, etc. and with minimal information about the practices attached to these forms of indexing, which would have indicated just how common and widespread they were, down to the level of local notaries and village tax collectors.[17] As a result of this short shrift, core Roman documentary practices illustrated by hundreds (not dozens!) of papyri are dwarfed by a long and speculative discussion of practices scantily documented by Latin inscriptions (e.g. the organization of amphitheater seating) in which Riggsby concludes that nested lists were extremely rare in the Roman world by virtue of being a complex operation with significant cognitive demands (38). This is a false conclusion that dramatically misrepresents the evidence. It remains a mystery why Riggsby continues to refer to these well-documented Roman listing techniques as “standard within certain narrow contexts” but not “common in the world at large.” Anyone who notarized a sale, authenticated a will, or obtained a receipt for paying taxes or filing a census declaration in the Roman empire—and this covers the majority of the imperial population—would have come into direct contact with the practices of indexed and nested listing.
As examples of evidence consciously excluded, Riggsby states that he has deliberately omitted “a variety of institutionally important contexts such as the census since they don’t happen to provide evidence that addresses my specific claims.” As noted above, the census offers a clear illustration of indexed and nested listing as core Roman documentary practices, so it is hard to see why it was excluded as irrelevant. It is similarly difficult to understand why dozens of papyrological examples of alphabetical ordering (e.g. alphabetical lists of landowners and taxpayers)[18] and registers of names with column numbers that enable entries for specific persons to be located in rolls of administrative records[19]—evidence that directly pertains to the subjects of ordered lists (11-15) and tables of contents (22-29)—should have been deliberately omitted.
This brings us to a general issue of historical methodology. To identify a set of techniques for managing information[20] as “advanced” and apply them as a kind of litmus test to sources from the Roman empire may already be questioned as arbitrary. To then define some of these techniques very narrowly (e.g. multicolumn tables to be read in both directions), generating a negative discussion of their absence, while excluding large bodies of evidence illustrating others, is deeply misleading. And it is the supposed absence or scarcity of these “advanced” technologies in the Roman context that underpins Riggsby’s overarching argument that the Roman informational landscape was balkanized and primitive. Riggsby himself criticizes this sort of approach in primitivist scholarship on Roman art (133-135), rightly emphasizing that “if we define Roman art simply by what it (supposedly) lacks, we will fail to attend to how it actually operates” (135). One only wishes that the same principle had been observed in the argumentation of this book.
It is certainly true, as the author states, that the conclusions of the book are “straight-forwardly falsifiable,” and I have indicated a number of contexts in which the evidence falsifies these conclusions.
Notes
[1] In addition to the alimenta tablets (CIL IX 1455 and 1147) I cited the example of BGU I 95 (Arsinoite, 147 CE), a census declaration where a bare reference to a “house and courtyard” constitutes an obligatory cross-reference to property registers, and P.Lond. III 1164e (Antinoupolis, 212 CE), a contract of sale containing detailed information about a property that illustrates the sort of data that census declarations omit. Contrary to Riggsby, P.Lond. III 1164e was not cited as an example of an obligatory cross reference.
[2] See e.g. PSI V 450 (Oxyrhynchus, 2-3c. CE) verso 1,48-68 for an extract from the files of a register of real property, ordered by location, gender and name, and 2,69-88 for an extract from records of topographical inspection verifying the identity of residents in each house of each urban district, likewise ordered by location and name.
[3] This is a paraphrase of BGU I 118,2,3-8 (Arsinoite, 189 CE). See Riggsby 2019: 20: “there is a strong Roman norm against what I’ll call “obligatory cross-reference. They do not like to force readers to look up a second text to be able to understand what they are reading in the first.”
[4] See e.g. Duncan-Jones 1976b: 56: “these two sources (metrological writer and Epiphanius) thus agree in implying that the modius castrensis was the modius of 1 1/2 modii Italici whose existence is widely attested under other names in metrological writers and papyri;” 57: “the second (Ptolemaic) artaba, that of 4 1/2 modii Italici, points to a modius castrensis of 1 3/8 (1.375) modii Italici, the precise amount of the modius xystos defined as 22 sextarii. This interpretation must be accepted, since the chances of an exact equivalence based on two known measures being thrown up randomly by the co-ordinates of some other calculation are virtually negligible.”
[5] See C. Corti, “Pesi e misure nei commerci, arti, mestieri e professioni,” in C. Corti and N. Giordani (eds.) Pondera: pesi e misure nell’antichità, Modena 2001, 143-166 at 144-145 and “Le misure di capacità,” ibid. 219-226 at 222-223; Mommsen ed. CIL III 14165 at 2316.
[6] The exactum ad Castoris weights are catalogued by F. Luciani and T. Lucchelli, “Pondera exacta ad Castoris,” Instrumenta inscripta VI. Le iscrizioni con funzione didascalico-esplicativa, Trieste 2016, 265-289. Beyond four photos, no information is given about the condition of the specimens. From 71 weights where the mass is specified, Riggsby chooses a sample of 55 and compares them to the 327g standard of the Roman pound, arriving at the following figures: “38 are off by at least 2%, 18 by at least 5%, 10 by at least 10% and 8 by 20% or more with a maximum deviation of 53%.” It is utterly unclear to me how Riggsby’s calculation misses the fact that 23 of the weights deviate by less than 2%. Riggsby arrives at discrepancies of 10 and 20% by doubling deviations of 5% and 10% to create a range above and below the standard. This estimate is flawed, however, because an overwhelming majority of surviving Roman weights are underweight—through use-wear, damage, and as a general pattern, see the detailed discussion in RIB II 2412—and only rarely overweight, typically by small amounts.
[7] For example, the Feldkirchen set (AE 2001, 1582) contains 9 weights underweight by only 0.1, 0.3, 0.3, 0.69, 1, 1.1, 1.3, 2.3 and 2.7% while the second-smallest weight (43.79gr) is underweight by 20% and the smallest (29.05gr) overweight by ca. 1.8g = 6.6%.
[8] On corrosion, which if internal may be invisible and accounts for up to 50% weight loss in coins with a high copper content, see K. Butcher and M. Ponting, The Metallurgy of Roman Silver Coinage: From the Reform of Nero to the Reform of Trajan, Cambridge 2014, 91-92.
[9] P. Grierson, “Weight and coinage,” Numismatic Chronicle and Journal of the Royal Numismatic Society 4 (1964) iii-xvii at ix, with examples of precisely calibrated stone weights listed at xii; see also RIB II 2412,86 (underweight by 0.34%) and R. Duncan Jones, Money and Government in the Roman empire, Cambridge 1994, 214 for a median average deviation of 0.6% in two samples of stone weights from Naples (CIL X 9067). Riggsby’s remark that these weights exhibit “somewhat less variation than in other samples” (102) elides these figures. On glass coin weights, see e.g. C. J. S. Entwhistle, “Byzantine weights,” in A. Laiou (ed.), The Economic History of Byzantium:
From the Seventh through the Fifteenth Century, Washington 2002, 611-614.
[10] On the metrology of the Roman denarius, see the standard reference work of D. R. Walker, The Metrology of the Roman Silver Coinage (3 vols.), Oxford 1976-1978, and the more recent treatment of Butcher and Ponting 2014 (n. 8) 155-460. Neither work is cited by Riggsby. On the calculation of the Roman pound from the evidence of coin hoards, see J. Elsen, “Le système podéral romano-Byzantin. Fin 3e siècle – fin 8e siècle,” Bulletin du Cercle d’études numismatiques 42 (2005) 101-113. Contrary to Riggsby, the variability of local standards for bronze coinage does not cast doubt on the standardized production of Roman denarii (96). Naturally, imperial standards for precious metal coinage were subject to reform and coins struck on different standards could circulate simultaneously.
[11] See AE 1935, 50 and 194 (Philippi, Macedonia, imperial period); see further CIL IX 2854 = ILS 5591 (Histonium, Italy, 1-2c. CE). These are cited by Riggsby in a footnote (111 n.85) but not discussed. For the recalibration of weights, see e.g. R. Reich, “New insights on the function of the Jerusalem agoranomos (market inspector) in the early Roman (= Late Second Temple) period,” in B. Jürgen, T. Laubach and K. Lindner (eds.), Zeichenlandschaften: Religiöse Semiotisierungen im interdisziplinären Diskurs, Bamberg 2021, 337-353.
[12] On the dual prosecution of forgery civiliter and criminaliter, see the detailed discussion of S. Schiavo, Il falso documentale tra prevenzione e repressione: impositio fidei criminaliter agere civiliter agere, Milano 2007 (not cited by Riggsby).
[13] See also Riggsby 2019: 199: “the one known appeal to this archive failed” with reference to CIL X 7852, which is misleading because the consultation of the archive at Rome coincided with the civil wars and urban fires of the year 68. For other examples of “authority” used with reference to knowledge claims not corresponding to practical capabilities, see Riggsby 2019: 20 (Cicero quoting a legal text by chapter that he proceeds to paraphrase) and 191 (Agrippa’s map as a gesture of authority with no administrative application).
[14] P. Tebt. II 287,5-9, 161-169 CE.
[15] See Phlegon, Peri makrobion FGrH 257 F 37; for records organized by birth year, see e.g. P.Mich. XI 603 (Arsinoite, 134 CE).
[16] See P.Oxy. II 237,7, 21-27; the Flavian edict is cited verbatim in the following lines.
[17] See e.g. P.Mich XI 625 (Arsinoite, 121 CE), a receipt for the payment of a tax on a loan kept by the payer, specifying the payment’s location in the records of the bank, organized as a nested list; P.Oxy.Hels. 35 (151 CE), a copy of a deed of divorce from a roll of notarial records, organized as a nested list.
[18] See e.g. P.Col. II 1 recto 2 (Arsinoite, 129 CE); P.Col. V 1 verso 3 (Arsinoite, 155 CE).
[19] See e.g. P.Mich. VI 380 (Arsinoite, 2c. CE) and P.Erl. 49 (Arsinoite, 3c. CE).
[20] Alphabetized lists, tables of contents, indexed lists, nested lists, “obligatory cross-reference,” multicolumn tables.