BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 The Beginnings of the Roman Census in Egypt Roger S. Bagnall p\:.0MAN PROVINCIAL CENSUS in Egypt, conducted at four­ teen-year intervals from at least A.D. 33134 to 2571258, has been known since the beginning of this century, and published declarations represent every census during that span. The earliest and latest years are the least well documented, but sufficient evidence demonstrates that census declarations were submitted for the years 33/34, 47/48, 61/62, and all years at fourteen-year intervals thereafter through at least 2571258. 1 Nor are any surviving declarations attributable to any year outside the fourteen-year cycle later than 33134. The scarcity and ambi­ guity of earlier direct evidence, however, has generated an ex­ tensive controversy during the past six decades about the dates at which the Roman government instituted the census in Egypt and at which the census began to be conducted on the fourteen­ year cycle. With the discovery of new evidence, it is now possible to show definitively that although there were census declarations before 19, the fourteen-year cycle cannot have be­ gun before that year and may even be later. On the other hand, some evidence suggests-without quite demonstrating-an earlier seven-year cycle dating back as far as 11110 B.C. The Problem The small number of surviving declarations has meant that it is not possible even to demonstrate that the cycle was intended as early as 33134. The coincidence of the fourteen-year intervals 1 The discussion of M. HOMBERT and C. PREAUX, Recherches sur Ie recense­ ment dans I'Egypte romaine (=Pap.Lugd.Bat. 5 [Leiden 1952]: hereafter 'Hombert/Preaux') 47-53, is still the fundamental treatment of the fourteen­ year cycle. A revised list of declarations published up to about 1973, prepared by G. Nachtergael, appears in P.Brux. I pp.51-58. A complete list with critical discussion will appear in R. S. Bagnall and B. W. frier, The Demography of Roman Egypt, forthcoming. 255 BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 256 BEGINNINGS OF THE ROMAN CENSUS IN EGYPT makes it likely at the least that 61/62 was chosen because the interval since 47/48 was the same as that between 33/34 and 47/48. That, in turn, suggests no full censuses between 33/34 and 47/48, or the interval would not have been meaningful. That view conforms to the silence of the declarations. The registers of taxpayers were periodically updated during the interval be­ tween those two years, but no evidence attests that taxpayers were asked to submit new declarations during that intervaI.2 None of this, however, tells us anything about official inten­ tions in the 30s and 40s. The pre-33 evidence is very obscure. The earliest texts with any resemblance to census declarations arc documen ts of 19 and 18 B.C. in which a royal farmer from Theadelphia, one Pnepheros son of Phancmieus, reports himself in what he describes as a U1tOIlVTlI.W (W.ehr. 200=P.Grenf 145-46). Pncph­ eros describes himself as 63 in the first declaration and 64 in the second; he does not include any other members of his house­ hold in his report, and no external evidence tells us if any existed. These apparently annual reports are of uncertain purpose and have neither predecessors nor successors.3 They may represent an early attempt at a generalized census, but the absence of other family members suggests that self-declaration as a royal farmer may have been the object (as S. L. Wallace thought).4 The center of controversy, however, has been another text. P.Mil. I 3 (PLATE 1a), since 1928 universally regarded as probably the first surviving declaration, was filed by a fifty-five­ year-old public farmer and priest named Harthotes son of Marres, who declares his son and mother along with himself. The papyrus is broken at the bottom in the middle of the oath formula and before the date. The oath is by Kaicrapa, which is the normal beginning of oaths by Augustus. s The editor, A. 2 This whole question will be treated by A. E. Hansen in her forthcoming work on The First-Century A.D. Tax Archi7)e from Philadelphia, where she will reedit P. Mich. X 578. See also preliminarily her remarks in P. Congr. XV 13.6n. (pp.65f). 3 See Hombert and Preaux 51 f on this matter. • "'Census and Poll-Tax under the Ptolemies," AJP 59 (1938) 432. Curiously enough, Wallace does not cite these papyri in his Taxation in Egypt (Prin­ ceton 1938). 5 E. Seidl, Der Eid im romisch-agyptischen Pro7)inzi.drecht I (=MunchBeitr 17 [Munich 1933]) 11, 18H. An additional example has since appeared in P.Amst. I 28. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 ROGER S. BAGNALL 257 Calderini, cited parallels for the oath formula only from the reign of Augustus, but he took no account of these in his discussion of the date, in which he left open the choice of 19/20 or 33/34. In his book on oaths, Seidl, assuming that the fourteen­ year cycle had begun already under Augustus, took the date of the papyrus, with its oath by Augustus, to be A.D. 5/6.6 This date might have won acceptance had not the publication, while Seidl's book was in press, of P. Oslo II 32, from A.D. 1, presented scholars with further evidence for the declarant. Harthotes son of Marres was in this papyrus stated to be 40. In 5/6, of course, he would have been 44, not 55. The editor of the Oslo papyrus, S. Eitrem, pointed out the anomaly and suggested a date in A.D. 16, accepting the evidence of his new text; a declaration at such a date would indicate that the fourteen-year cycle was not yet in place. Calderini essentially accepted this view in a subsequent article, but expressed doubts about the accuracy of the age indications in the Oslo papyrus and cautiously left any conclu­ sion to the future and the appearance of new evidence.? Hombert and Preaux (50), however, pointed out that the age given in the contract of A.D. 1 had the appearance of a rounded figure and need not be taken as precise. They offered no date of their own for P.Mil. 13. The controversy was relaunched by the republication of P. Mil. I 3 by S. Daris, in the second edition of the Milan papyri (1967). Daris pointed out the problem posed by the oath for­ mula and the Oslo papyrus, as a result of which he left open the date. In a review of Daris' work, John Rea proposed taking the oath formula to refer to Tiberius, allowing a date of 19/20, in which case there would be a three-year discrepancy in Har­ thotes' ages, readily explicable by age-rounding. 8 Rea's hypoth­ esis also would explain why Harthotes' daughter Taphaunes, born ca 1 B.C. and thus 38 in A.D. 38 (P.Mil. I 7), is not mentioned in the declaration: she is already married and living in her hus­ band's household. Since she had a son Aunes, described in A.D. 38 as 22 years old (P.Mil. I 7), in A.D. 44 as 25 (SB XIV 11279), and in A.D. 47/48 as 28 (P.Mil. I 8), thus presumably born ca 16- 6 Seidl (supra n.5) 47, 136 (BL 3.117). 7 "La piu antica schcda di ccnsimcnto Romano provenientc dell' Arsinoite," RendIstLomb 64 (1931) 551-58. "Altre ricerche 0 altri documenti che i papiri ci rivciino potranno dichiarare meglio la cosa dimostrarlo vera 0 chiarire per altro mezzo la reald dei fani," he concluded. 8 CR N.S. 19 (1969) 95 (BL 7.101). BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 258 BEGINNINGS OF THE ROMAN CENSUS IN EGYPT 19, that is very plausible. 9 But Rea's proposal requires us to assume an otherwise unattested blunder in the oath formula, and only a presupposition that the census was already at this time on a fourteen-year cycle could justify such violence. When Orsolina Montevecchi took up the question again in 1976, these considerations led her to look to the latter part of Augustus' reign for a date for this declaration. Pointing to the reference in P.Oxy. II 288.35 to an £7tllCptCnc; of the forty-first year (11112), which was the basis of a list of persons with their ages, and a succeeding reference to £~ c11t[o'Ypaq)'TlC; lCro]Jlo'YpaJl­ Jla't£rov Jl~ (£'to'UC;), she proposed that a census was held in 11112, with declarations due in 12/13, and that the Milan dec­ laration refers to this census. 10 Hombert and Preaux (49) had already suggested the likelihood of a census in 11/12, but they had not drawn the conclusion that P.Mil. I 3 belonged to that year. Montevecchi thus took epikrisis to refer to the census. 11 There is, however, no reason to think that at this time declara­ tions were filed in the year following the census as was later the case. 12 One could therefore take the phraseology of P.Oxy. II 288 in the most straightforward manner, as pointing to two processes, an epikrisis in 11112 and declarations in 12/13. Now C. A. Nelson has argued that the first known general status epikrisis in Roman Egypt took place in 4/5,13 and points out that 9 Cf G. Casanova, Aegyptus 55 (1975) 129. This sequence gives a good indication of the inexactitude of age indications in contracts. 10 O. Montevecchi, "II censimento romano. Precisazioni," Ae'Vum 50 (1976) 72-84 at nf. 11 Following Hombert and Prcaux 48f; the question is dealt with in detail ~y H. Braunert, "Zur Terminologie der Volkszahlung im friihen romischen Agypten," in Symbolae R. Taubenschlag dedicatae III (=Eos 48.3 [1956: Bra­ tislava 1957J) 53-66, who concludes similarly that "in der friihen Prinzi­ patszeit wird die Volkszahlung als £1tlKPtOt<; bezeichnet" (56). 12 No regularity can be observed before 89 in the date of filing of returns. All three of those for 89 were filed in 90/91, but for earlier censuses no such pattern is found. For the previous census, P.Oxy. II 361 was apparently filed in 76/77 (I have seen a photograph), but the year number in the date is uncertain. For the two Arsinoite declarations for 75, B G U XI 2088 and P.Harr. I 70, there is no indication that the census year is past, nor is the exact date preserved. Both declarations for 61 were filed during 61162. The only declaration for 47 was filed early in 48/49. P.Oxy.Hels. 10, for 33/34, was filed during 33/34, as were the two Arsinoite declarations for that year, SB X 10759 and I 5661. 13 A register compiled in the 34 th year of Augustus (4/5) is mentioned in several Oxyrhynchite documents, cf Aegyptus 54 (1974) 29 and see below. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 ROGER S. BAGNALL 259 this would be the year preceding a putative census in 5/6.14 The next such benchmark general epikrisis seems to have been carried out in year 7 of Nero, 60/61, the year preceding the gen­ eral census of 61/62.15 The epikrisis, establishing privileged statuses, would be held reasonably enough before a general registration was required. But that hypothesis demands a regularity, a distinction between epikrisis and declaration, and a respect for technical terminology that cannot be demonstrated for the time of Augustus, and it also embodies a desire for cyclical regularity dovetailing into A.D. 19. At this point a discussion of the text commonly assigned to the census of 19 is necessary. This declaration, P.Oxy. II 254, preserves neither any reference to a year of the census nor a date for the declaration. Grenfell and Hunt, who supported an early introduction of the fourteen-year cycle, dated this dec­ laration to the census of 19120 on the basis of its address to Eutychides and Theon, topogrammateis and komogrammateis, a pair known to them also from P.Oxy. II 252, dated to 19120, a notice of the anachoresis without property of the declarant's brother. The same pair, but with the names in reverse order, are the addressees of P.M ich. X 580, a similar declaration of ana­ choresis in the [ ... J-rou year of Tiberius. The editor refrained from restoring the numeral (third, fourth, fifth, sixth, ninth, and tenth are possible), evidently on the grounds that these offices had a five-year term. The length of term is in fact unknown, but a minimum of six years is attested in one and perhaps both of the only other known pairs of officehoiders, all Oxyrhynchite. 16 14 C. A. Nelson, Status Declarations in Roman Egypt (=Am.Stud.Pap. 19 [Amsterdam 1979]) 23 with n.70. On no hypothesis would there have been an Egyptian census in 6/7, the probable year to which Luke credits that carried out in Judaea by Quirinius; but since Quirinius' arrival marked the imposition of direct Roman rule, such a non-coincidence is hardly surprising. See 11. Braunert, "Der r6mische Provinzialzensus und der Schatzungsbericht des Lukas-Evangeliums," Historia 6 (1957) 192-214. 15 Cf e.g. W.Chr. 147 (94/95), in which the declarant refers to his parents' marriage before year 7 of Nero and his mother's registration in the census of the next eighth year (lines 25-28). 16 Apollonios and Didymos are known in P.Oxy. XXXIII 2669 (41-54) and P.Mich. III 170 (49); if the Didymos and [] known in P.Oxy. II 251 (44) and 255 (48) are the same, a span of at least 44 through 49 would be attested. Apollophanes and Diogenes are attested at least from 62 (SB XII 10788B) through 63/64 (P.Gen. II 94) and 65 (PS! VIII 871) to a date that at earliest can be 66/67 and may well be as late as 72173 (P.Geno'V. 112, cf Youtie, ZPE 24 [1977J 138f= Scr. Post. I 400f), since it involves a royal scribe whose (appar­ ent) predecessor is attested through 65/66 and who is himself attested in 72173. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 260 BEGINNINGS OF THE ROMAN CENSUS IN EGYPT A date for P.Oxy. II 254 in 19/20, then, is an economical hypothesis, but not a necessary one, since a date within six years, or even more, on either side of that date would also be acceptableY Thus despite the lack of secure evidence for a census in AD. 19, it remains attractive on the basis of P.Oxy. II 252 to date P.Oxy. II 254 to or near that year. This evidence would not seem to show more than indications of government attempts to establish population lists around 4/5 and 11112, with years 34 and 41 acquiring a certain benchmark status justifying their later citation. 18 New Evidence Only one month after I had written a draft of the preceding pages, concluding that the most likely (or at least most at­ tractive) hypothesis was a pair of seven-year intervals before A.D. 19 and a date in 12/13 for P. Mil. I 3 (like Montevecchi, though for different reasons), I had the extraordinary fortune to find the missing lower half of the papyrus while browsing in the Columbia papyrus collection (PLATE 1 b). With the inventory number 8,19 it is a near neighbor to inv. 6, published in 1928 by C. W. Keyes and now republished as P. Col. VIII 209. 20 P.Col. inv. 6 is a petition to the basilikos grammatetts by Marsisouchos, the brother of the Harthotes of P.Mil. I 3, dated A.D. 3. It was noticed and its significance seen by Eitrem in his commentary 17 G. M. Browne, in the introduction to P.Mich. X 578, argued that the census list published there referred to a census in 19/20, thus strengthening the argument for dating P.Oxy. II 254 to that year. Hanson's reedition (supra n.2) will present an argument that the list in fact refers to the census of 47/48; but removing that prop to the argument does not help to establish any particular date for the Oxyrhynchos declaration. 18 Braunert (supra n.11) 65 argued that the census was still annual as late as 12/13. His basis was P.Oxy. II 288 and 314, in which declarations of year 42 are mentioned. Since then P.Oxy. II 314 has been published in full (Aegyptus 46 [1966] 77 [SB X 10220]); the ages are given as in year 42, and at the foot is stated 'tE't£A(EeJ'tat) Il~ (£'touc;). This seems to me compatible with a two-part process and not to require the hypothesis of annual declarations. 19 Described in Bell's inventory simply as a hypomnema. 20 See the introduction there for other versions of this petition. Inv. 7, unpublished, also belongs to the archive. It is a loan dated 9/10 in which Harthotes is the borrower. It is difficult to repress surprise that inv. 6 was published without anyone's noticing that the next two items in the collection were connected. Inv. 7 will be included in P.CoL XI. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 ROGER S. BAGNALL 261 to P. Oslo II 32. Both Columbia papyri were purchased in 1923 through the British Museum consortium, as was the Oslo papyrus. Dated to year 41 of Augustus, Tybi [?2]6, or 22 January AD. 12, the Columbia fragment justifies fully the view of Hombert and Pn!aux and Montevecchi that there was a census in 11112. Since a considerable number of improved readings in the first part of the text have appeared since Daris' publication,21 I give the complete text of the papyrus here. Lines 1-13 belong to the Milan papyrus, lines 15-27 to P.Co/. inv. 8; they share line 14, and a vertical marker has been put there to show the point of juncture (see PLATE 1). 'ICH()WPCP KO)f.lOYPUf.lu'tt 8E­ U()EACPf)U<; 1tUpa 'Ap8w"tOu 'tou Mupl1ou<; ()lWT is ]crl<; yEWp- 4 1'0<; KUt dEP£1)<; To8of)ou<; 8EQu. £xw £V 8EU()EACPf)u ohduv . £v'to<; 1tEPl~OAoU ~[l]EpOU, £V n £yoo uu'to<; 'Ap8w'tT\<; f.lT\'tpo[<;] 8 'EcrEpcru8EQ<; {(£'twv)} 1tEV['tf)]KOV'tU 1tElv]­ 'tE, 'Ap1tu'to8of)ou<; u\,o<; (hwv) EVV£U f.lT\tpo<; Tauvxopt- cplO<; KUt i1 f.lf)'tT\'p ·f.lOU 'EcrEpcru8[t]'<;' 12 ITucrtwvo<; £TWV {E'tWV} o. 'Ap8w't[T\]<; W 1t PO)YEYPUf.l£VO [<;] Wf.l vUO) Kutcrulp'u [A]U['t]OKpa't[oJpu 'EAEU8Ep(lOV) 8EQU Ma LE~Ucr't'ov i f.ll)v ... 16 E~ UylOU<; KUt E1t' apT\8Eu~ £1tl­ ()E()O)KEVUl 'to 1tpO~tf.lEVOV U1tof.lvT\f.lU f.lT\()£V U1tocr'tEl­ Aaf.lEvo<;. EuopKouyn ~£V 20 f.lOt d) tT\. £~lOpKouvn ()E 'tex £v­ UV'tlU. 'Ap8w'tT\<; Mupf)ou<; W 1tPo)),£- YPUf.lEVO<; KL (£'twv) VE OU(Al) CPUKCP f.lf)- 24 pcp aplcr'tEPCP. 21 Corrections by J. Bingen and D. Hagedorn, reported in BL 6.75, and M. Vandoni in BL 7.101. Some were merely recovering the correct readings given by Calderini in the first edition. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 262 BEGINNINGS OF THE ROMAN CENSUS IN EGYPT 'AP1tu'toSoilc; C;L WC; (hrov) S. KU'tUKEXWp(l(HUl) (£'tOUC;) f.1U Ku£aupoc; T~~[l ?K]C;. 1-2 Kroflo'Ypa) .. q..l.a:t£t eW~EAS ~llflooio1> 'YErop'YOU Kat iEP£roS 5 eEa~EAS (, npo'YE'Ypaflfl£VOS 23-24 fl TtAcr To Isidoros, village secretary of Theadelphia, from Hartho­ tes son of Marres, public farmer and priest of the god Tothoes. I have in Theadelphia a house inside the enclosure of the temple, in which (live) I myself, Harthotes, whose mother is Esersythis, fifty-five years old, my son Harpatothoes, nine years, whose mother is Taanchoriphis, and my mother Esersythis daughter of Pasion, seventy years. I, Harthotes, the aforewritten, swear verily by Caesar Imperator Eleutherios, son of the god, Zeus Augustus that I have submitted the previous declaration salutarily and truthfully, reserving no­ thing; if I have sworn truthfully may it be well with me, if falsely, the opposite. Harthotes son of Marres, the aforewritten, twentieth year (?), fifty-five years, with a mole on his left cheek. Harpatothoes, sixth year (?), about nine years. Registered in year 41 of Caesar, Tybi [?2]6. 1 Isidoros the komogrammateus is attested only here. 2 For Harthotes, his family, and their documentation, see G. Casanova, "Theadelphia e l'archivio di Harthotes," Aegyptus 55 (1975) 70-158. If the age of fifty-five here is exact, he was forty­ four in A.D. 1 when he is said to be forty by P. Oslo II 32. 14-15 The normal word order of this oath is Ku£aupu AU'tOKPCX'tOPU SEQU UtOV ~£u 'EAEUS£plOV IE~ua'tov (supra n.6, P.Mil. I 7); the scribe's problems with it here are extensive. 18 It is notable that the writer uses U1t0f.1Vllf.1U, the term found in W. Chr. 200 (cf above). I have found no parallel in the papyri for f.1ll()£V U1tOa'tElAaf.1EVOC; in this context, although an oath f.1 11 U1tEa'ta[AS(ul) in P. Oxy. II 246.26 is close. The phrase is, however, classical (cf LS]9 s.v. U1tOa't£AAw II.4) and occurs in P. Yale I 42.22, where the writer exhorts his correspondent not to hold anything back. BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 BAGNALL PLATE 1 (a) P. Mil. I 3, courtesy Universita del Sacro Cuore (b) P. Col. inv. 8, courtesy Columbia University Libraries, Rare Book and Manuscript Library (reduced to 70%) (a) (b) BAGNALL, ROGER S., The Beginnings of the Roman Census in Egypt , Greek, Roman and Byzantine Studies, 32:3 (1991:Autumn) p.255 ROGER S. BAGNALL 263 23 ou(A.:r,) seems superfluous, since the mole (