^{page 1 note 2} Anth. Palat. 11. 442.

^{page 1 note 3} For tables of reconstructions by nineteenth and twentieth century scholars of the ‘true reading’ of the papyrus of Athenaion Politeia cf. the edition by Sandys (p. 58) where already in 1893 a table could be produced of variants. The papyrus had been discovered only two years earlier! For the stage of scholarship by 1937 see the table in Schachermeyr, F. ‘Peisistratos’ R.E. xix (1937), 171–2.Google Scholar

^{page 2 note 1} Amongst these sceptics are Cadoux, T. J., ‘The Athenian Archons from Kreon to Hypsichides’, lxviii (1948), 104–9;Google Scholar P. Oliva, Studio Antigua (Prague, 1955), 2530 and Řana Řecka Tyrannis, 324 ff.; H. Bengtson, Griechische Geschichte (1960), 133; H. Berve, Die Tyrannis bei den Griechen (1967), 47 543 ff.; V. Ehrenberg, From Solon to Socrates (1967), 400 n. 12; Schreiner, J. H., ‘Aristotle and Perikles’, Symbolae Osloenses Suppl. Fasc. xxi (1968), 36 ff. (I owe this last reference to Prof. Ehrenberg). Partial sceptics who nevertheless believed that the first exile must have been of 4 years duration (see below) were Kenyon and Th. Reinach. See Aristotle on the Constitution of Athens, (ed. Kenyon, 1892); U. V. Wilamowitz, Aristot. and Athen (1893), 1 ff. More recently this view has been taken up by F. Heidbüchel (Philologus, ci [1957], 70 ff.) and G. V. Sumner (C.Q. xi [1961], 37–49) who believes that he has restored the original state of Aristotle's text, if not the history behind it. The theory of Beloch (Gr. G. i. 2. 289 ff.) was that Pisistratus had one exile and two spells of tyranny only. This was also Herschensohn's opinion (Philologicheskoye Obozreniye, x [1896], 119 ff.) and they were followed by Meyer (Forschungen …, 1899), De Sanctis (Atthis, 1912, 27 ff.) and Kahrstedt (‘Megaldes’ R.E. xv. 1. 125–6). The latest holders of this view were W. Otto (S.B. [Munich, 1924], final vol., P. 9) and U. Wilcken (Gr. G. ed. 5 [1943], PP. 95, 266) but Schachermeyr's comprehensive article of 1937 and Cornelius's monograph of 1924 seem to have disposed thoroughly of Beloch's ingenious, but unlikely, suggestion. This had also brought the tyranny of Theagenes of Megara and the attempt of Kylon at Athens into the years at the middle of the sixth century B.C. (Gr. G. i. 2. 302 ff.). P. N. Ure, The Origin of Tyranny (1922), also argued strongly against Beloch's view, but produced no set of dates for Pisistratus himself.Google Scholar

^{page 2 note 2} Mathieu, in his edition (Aristotle [1915], 29 ff.) tended to a literal acceptance of the numerals in the text, and was followed in this by Lang, M. (A.J.P. lxxv [1954], 59–73)Google Scholar, who consequently had to make the first coup of Pisistratus as early as 566 B.c. In this she was following a hint in C. Seltman, Athens, its History and Coinage (1924), 36 ff., to the effect that Pisistratus may have been behind the first setting up of the Panathenaic Games in the archonship of Hippokleides (a Philaid) in 566 B.C. However, this suggestion has found no favour. M. Miller (Klio, 1959) follows in this tradition with some refinements (see below). Various emendations for the numerals concerning the second tyranny and the first exile have been offered by scholars who accept the A.P. numeral for the first tyranny. Bauer (Forschung. zu Aristot., [1891], 50 ff.) set off this trend. Busolt, (Gr. G. [1895], 317 n. 4), Bury (History of Greece, 1912 and 1951) and F. Cornelius were in this tradition (Die Tyrannis. in Athen [1929], 5 ff.). Schachermeyr, in his great survey of the literature, argues for this kind of solution, with certain indebtedness to the ideas of Pomtow. Finally Pomtow (Rhein. Mus. Ii [1896], 500 ff.), Hiller v. Gaertringen (I.G. i^2. 270 ff.), F. Adcock and F. Jacoby, Atthis 1949, 188–196, are in a tradition which the present writer wishes to join on this question. See also now J. K. Davies, Athenian Propertied Families (Oxford, 1971, ), 444–5. Pomtow suggested that the unit in the case of the first and second tyrannies should be changed to ‘months’ from ‘years’. This group of scholars has normally assumed that the first exile lasted into the fourth or third year . The whole argument has been made desperately complicated by the uncertainty whether Herodotus and Aristotle were using single inclusive or double inclusive reckoning in their ordinal numeration, cf. Schachermeyr, loc. cit.Google Scholar

^{page 3 note 1} The Athenaion Politeia has recently been treated as by Aristotle himself (Day, J. and Chambers, M., Aristotle's History of Athenian Democracy [1967], 1–24). It will henceforth be abbreviated to A.P. Day and Chambers do not deal with Pisistratid chronology, but do touch on other economic and political aspects of the tyranny (90–7, 175–6). For my purpose it does not much matter whether A.P. is by Aristotle himself or by a ‘research student’ of his.Google Scholar

^{page 3 note 2} Isocrates 16. 26 … Forty here may be a mere round number, but it could also refer to the period 547–508/7 B.C. (Raubitschek, , Rh. Mus. xcviii [1955], 258–62). Andocides in De Mysteries 106 does not give any duration for the tyranny of the Pisistratids, although his was probably the same eupatrid tradition as that of Isocrates. He merely talks of a victory of his ancestors at ‘palleion’. The only Battle at Pallene of which we know is that won decisively by Pisistratus. It may be that Andocides is covering up for his great-grandfather, but, as MacDowell points out, the generations would fit better with the period elapsed by Andocides' time if a return of the nobles along with the Spartans in 510 is being referred to. MacDowell disapproves of Raubitschek's explanation of the 40 years of tyranny, on the ground that Isagoras was not a tyrant, even if the Spartans wished to make him so in 508/7 B.C. (D. MacDowell, Andokides on the Mysteries [1962], Appendix O, 212–13). However, we can reply to this that Aristotle certainly could call Isagoras and his party ‘friends of the tyrants’. The Alcmaeonidae might well charge that the tyrant faction was not fully ‘out’ until Clisthenes had taken the people into partnership (A.P. 20-r). At the other end of the 40 years a case could be made out that the first two short periods of power (certainly the second one, held in partnership with Megacles) were not considered by certain noble families to have been periods of true tyranny. Certainly, Herodotus describes the events after the first coup in terms of a tyranny, but if it was very brief it would hardly seem worth bothering about after the Battle of Pallene which ushered in the real ‘tyrannis’ based on armed might.Google Scholar

^{page 3 note 3} The only support for this is in a figure given by Aristophanes of Byzantium, schol. on Aristophanes, Vespae 502, where the length of the tyranny is given as . This, however, could be a corruption of the Greek numeral for 49 or 51. Eratosthenes, quoted by the scholiast ad loc., seems to have reckoned the duration of the tyranny at 50 years, a tradition also followed in the Parian Marble, see below, n. 4. Justin 2. 8. to gives a figure for the rule of the Pisistratids of 34 years, which seems to be an inaccurate reflection of the 36 years of power mentioned by Herodotus and A.P. and the 35 years in Aristotle, Politics 1315^b.

^{page 3 note 4} The Parian Marble, which perhaps depends on Philochorus, the last of the Atthid historians, seems to have a base date of 263/2 B.C. Its calculations are: (a) Epoch 450 ‘297 years since Pisistratus became tyrant of Athens in the archonship of Komeas’ (560/59 B.C.). (b) Epoch 45. ‘248 years since Harmodius and Aristogiton killed Hipparchus, and the Athenians drove out the Pisistratids from the Pelargikon Fort in the archonship of Harpactides’ (511/10 B.c.). See F. Jacoby, Das Marmor Parium (Berlin, 1904). In spite of the authority of Herodotus, Thucydides and Aristotle that Athens was freed in 511/10 B.C. by the Spartans at the request of Delphi and the Alcmaeonids, and that Hipparchus was murdered in the fourth year before that, the Parian Marble has succumbed to the vulgar Athenian version against which Herodotus (5. 55 and 123) and Thucydides (1. 20; 6. 54–9) had polemized. Its account, at epoch 45 at least, is unreliable.

^{page 4 note 1} Eusebius (Armenian version, ed. Schoene, 94, 96, 98, 100); Jerome, ed. Fotheringham, 181, 183, 185, 188. Four entries are given in each version. (a) ‘Pisistratus apud Athenicnses tyrannidem exercuit et in Italiam migravit’ (Arm.); ‘Pisistratus Atheniensium tyrannus in Italiam transgreditur’ (Jerome). The name Italia appears to be a corruption of Thracia to which Pisistratus repaired during his later second exile. Eusebius dated this contemporary with the last year of Alyattes or the first years of Croesus. Probably he intended c. 561/60 or 560/59 which can be regarded as the most probable date of the first coup. (b) ‘Pisistratus Atheniensium iterum imperavit’ (Arm.); ‘Pisistratus secunda vice Athenis regnat’ ( Jerome). This return of Pisistratus is variously dated in the manuscripts from C. 547 to 542/1 B.C. It is to be noticed that Eusebius implies for Pisistratus only one exile, and two periods of tyranny. It was on the strength of this fact, along with a version told by the unreliable Polyaenus (Strat. 1.21. 1) who placed the ‘Phye’ episode during the Pallene campaign, that Beloch and Herschensohn thought that the two exiles and returns of Pisistratus were mere duplicates. We must, however, follow Herodotus and Aristotle in this matter, and they explicitly state the circumstances of the two exiles which are quite different. (c) ‘Iparchus et Ipias Atheniensium tyranni cognoscebantur’ (Arm.); ‘Hipparchus et Hippias Athenis tyrannidem exercent’ (Jerome). The manuscripts again vary in allotting this event to various years between 529 and 521 B.C. with a preponderance in favour of c. 528/7 B.C. This year can be worked out independently of the chronographers as that of the accession of the sons and is not a subject of dispute. (d) ‘Armodius et Aristogiton …’ (Arm.); ‘Harmodius et Aristogiton Hipparchum tyrannum interfecerunt et Leaena meretrix arnica eorum … (Jerome). The Hipparchus assassination was set in either 522 or 521 by the manuscripts. This event, however, can be fixed more reliably than almost any other sixth-century date, during the Panathenaea of 514/13 B.C. (Aug. 5 t 4). Obviously we can rely on neither the dates nor the information which is retailed by Eusebius, for exactitude.

^{page 4 note 2} Shear, T. L., ‘Three Women of Athens Koisyra’, Phoenix, xvii (1963), 104;Google Scholarden Boer, W., Mnemosyne, xx (1967), 57–8;Google ScholarEliot, C. W. J., Historia, xvi (1967), 283 n. 24. Miss White (in an unpublished address given at Leeds in 1965 to which my attention was drawn by Mr. R. Lock of Victoria University, Wellington, New Zealand, who was present and took notes) apparently still feels that the testimony of Herodotus is the best basis for considering the Pisisttatid period. This is my own view.Google Scholar

^{page 4 note 3} See Miller, M., ‘The Earlier Persian Dates in Herodotus’, Klio, xxxvii (1959), 41–5, esp. 42–3 for a table purporting to reconcile the numerals. Miller retains all the numerals in the text, as did Mathieu and Lang before her, but following Kaibel, Kiessling's edition of 1891, and Bauer, she attempts to save five or six years by taking to be the number of years elapsed since the first coup. This is an awkward, if not impossible, way of construing the Greek, since it should naturally refer to the ‘twelfth year’ from the intervening first expulsion. I argue below that Miller's scheme is open to the logical demand that the years of exile should total fourteen, since she makes the 36 years correspond to the total amount of all three tyrannies of Pisistratus plus that of the sons.Google Scholar

^{page 5 note 1} Adcock, F. E., ‘The Exiles of Pisistratus’, C.Q. xviii (1924), 174–81;CrossRefGoogle ScholarC.A.H. iv. 63 f. Adcock has been followed in this brilliant suggestion by almost all scholars in the English tradition until the recent articles mentioned above, e.g. C. Seltman, Athens, Its Early History and Coinage (1924), 43 n. 4; T. J. Cadoux, loc. cit. above, p. 2, n. 1; Hignett, C., History of the Athenian Constitution, (Oxford, 1952), 114; A. Andrewes, The Greek Tyrants, 100–1; A. French, The Growth of the Athenian Economy (1964), 31; W. G. Forrest, The Emergence of Greek Democracy, 180, and N. G. L. Hammond, History of Greece to 323 B.C. Miller has taken a retrograde step in thinking that the 36 years of Pisistratid tyranny covered the total period of power from 560/59 B.C. less the exiles.Google Scholar

^{page 5 note 2} Sumner, loc. cit. (above, p. 2, n. ), 37–48, esp. 38–41.

^{page 5 note 3} Sumner, ibid. 45–8. He emends the numeral for the first exile in the papyrus text to He acknowledges that he is following Wilamowitz and Heidbüchel. Kenyon in his third edition of 1893 also follows this suggestion.

^{page 6 note 1} For attempted explanations of the differences between the vulgar version of the assassination of Hipparchus, and the accounts by Herodotus, Thucydides, and A.P., Lang, M., ‘The Murder of Hipparchus’, Historia, iii (1954), 395–407;Google ScholarFornara, C., ‘The Tradition about the Murder of Hipparchus’, Historia, xvii (1968), 400–24.Google Scholar

^{page 6 note 2} Hammond, N. G. L., ‘Studies in Greek Chronology of the Sixth and Fifth centuries B.C.’, Historia, iv (1955), 393–6;Google Scholar‘The Philaids and the Chersonese’, C. Q. vi (1956), 113f.Google Scholar

^{page 6 note 3} F. Gr. Hist. 105 F (Jacoby). Hammond tries to show that the second expulsion occurred in the ephorate of Chilon and the kingship of Anaxandridas at Sparta (i.e. 556 B.c. according to traditional dating). The papyrus, however, seems to be too fragmentary to support the conclusion that it is an expulsion of Pisistratus which is meant, and not the much better-known and later flight of Hippias in which the Spartans certainly had a hand.

^{page 6 note 4} ‘La Chronologie de Pisistrate’, La Nouvelle Clio vii-ix (1955-1957)3 161–80.Google Scholar

^{page 7 note 1} La Tyrannie dans la Grèce antique (1969), 64.

^{page 7 note 2} Jacoby, F., Atthis, 1949, 188–96.Google Scholar In a recent short note on the exile of the Alcmaeonidae Bicknell, P. J. adheres to the scheme given by Jacoby, Historia, xix (1970), 129–30, as does J. K. Davies, Ath. Prop. Fam. 444. A similar line has recently been taken by P. J. Rhodes at a seminar held at Leeds University on 22 Nov. 1972. But as he accepts the emendation of the first exile to four years' duration (as the slightest emendation to produce consistency in A.P.), he leaves no time whatsoever for the first and second periods of tyranny. He also has no explanation for the ordinal numerals which appear in the text of A.P. In fact his reconstruction requires two numerals to be emended and two to be discounted altogether.Google Scholar

^{page 7 note 3} Fritz, K. von and Kapp, E., Constitution of Athens and Related Texts (1960), x-xi.Google Scholar

^{page 8 note 1} The use of to signify lapse of time in Greek narrative occurs in connection with both cardinal and ordinal numerals. By A.P. (13) the archon Damasias was said to have held on to office ‘for two years and two months’. Herodotus (4. 44) says of the fleet of Scylax sailing westwards around Arabia, It is clear that such and greater precision was retained in the tradition for some events in the sixth century. Herodotus speaks of the agreement of Hippias to withdraw from Athens within five days (5. 65). Sardis fell after a siege of fourteen days (I. 84). Thucydides (2. 21) gives a certain example of with an ordinal numeral for an event in the later fifth century, whether the numeral is correct or corrupt. Given the general practice that ordinal numerals were used with the unit of months, there seems no real reason why the Greek form should be considered unidiomatic.

^{page 9 note 1} F. Jacoby, F. Gr. Hist. III B, 328, F34. I owe this parallel to the reader of C.Q. who kindly drew my attention to it.

^{page 9 note 2} H. Pomtow, Rh. Mus. li (1896), 560 f.

^{page 9 note 3} The year 560/59 was thought to be the more likely by Miller. T. J. Cadoux prcferred 561/60, although he recognized tha 560/59 was possible, loc. cit. 104 f.

^{page 9 note 4} Herodotus I. 84–5. For recent attempt to fix the fall of Sardis, Kaletsch, H., ‘Zu Lydischen Chronologie’, Historia, vii (1958), esp. pp. 39–47Google Scholar; Miller, M., ‘The Herodoteai Croesus’, Klio, xliii (1963), 75–6Google Scholar, am ‘Herodotus as Chronographer’, Klio, xlvi (1965), 109–28Google Scholar. In the past it has been date in 546 or to 541 B.C. More lately the dat Oct./Nov. 547 B.C. has become fashionable Some scholars accept the Nabonidu Chronicle (discussed in the above-mentionei articles) as referring to the destruction c Lydian power by Cyrus. There is the diffi culty that the king of Lu ?—mentioned the Chronicle—is said to have been killer by Cyrus and he may have resided in direction distinct from Sardis, near Arbela For a list of a number of scholars wh have reservations concerning the reference c the Nabonidus Chronicle entry to the fall of Sardis, see Eliot, C. W. J., Historia, xvi (1967), 284Google Scholar n. 24. If Sardis fell in 546 B.C. then there is time for Croesus to discover the Athenians under a tyranny and the Spartans newly strengthened by victories over Tegea. The year 546 has the greater traditional support among the chronographers, and the only modern scholar who is inclined to assign the capture of Sardis to 541/0 B.C. 13 Mazzarino, S., Fra Oriente e Occidente (1947), 162–4.Google Scholar

^{page 11 note 1} Cadoux, loc. cit., 108, and Sumner, loc. cit., 47 n. 1, declined to believe that the archon Hegesias could be a corruption of Hegestratus, as was suggested by Pomtow, Rh. Mus. li (1895), 565, and by Jacoby, Atthis, 5949, 378 11. 543. Perhaps there were indeed two such archons but the A.P. might well have transferred events of the one year to the later year, because the author believed that the first tyranny lasted several years. It is unlikely that an expulsion decree dated by Hegesias' name survived the subsequent tyranny of Pisistratus to be used as evidence by the fourth-century B.C. writers.

^{page 12 note 1} Sumner, loc. cit. 46, comes to the opposite conclusion that the version given by Aristotle and A.P. was not a false interpretation as Heidbüchel suggested, loc. cit. 79, but ‘a deliberate reinterpretation’. Yet A.P. does not criticize the tale as told by Herodotus as we should expect of an ancient author. I rather follow Heidbüchel on this point.

^{page 13 note 1} The numerals are written in full in two clear hands on the papyrus which dates from c. A.D. 70–100. The better of the two scribes was still at work in sections 14–19. Of course, in the original text or in any of the intermediate copies, the numerals may have been in acrophonics and much more liable to become corrupt. Some ingenious emendations of the ordinal numerals depend on this latest supposition, although D. M. Lewis doubts that acrophonics were used for ordinals, Historia, xi (1953), 415. Jacoby suggested on palaeographic grounds the emendation ii for (i.e. for ). Thus his interpretation of the text and of the history behind it differs little from that put forward by me now since it is proposed to substitute ‘in the twelfth month’ for the return after the first exile. This is only a few months less than the version as Jacoby gives it. For favourable comments on the quality of the papyrus text, Jacoby, op. cit. 194; Sumner, 37–8.

^{page 13 note 2} Herodotus was in general no fanatic for chronology. He presents us with a mass of relative chronological material covering the whole Greek and much of the Near Eastern world, presenting king-lists, and correlations between contemporaries or successive generations. Yet he gives only one absolute date in the whole of his history (8. 51) where Xerxes crosses into Europe and four months later appears in Attica, This year, 480/79, appears to be a base line for Herodotus from which he worked backwards (7. 7; 7. 20. 1). Thucydides is also useful in the matter of the relative chronology of the fall of Hippias, the battles of Marathon and Salamis, and other sixth-century events such as the first alliance of Plataea with Athens, which took place probably during the Pisistratid tyranny in 519 B.C. (3. 68.5; 6. 59. 4). But Thucydides was also influenced by the new fashion for fixing events, particularly contemporary ones, by archon years (2. 2) and other Greek magistracies. He showed himself to be fully aware of the difficulties involved in working out the chronology even for the immediately preceding fifty years. He criticises Hellanicus (1. 97. 2) and himself puts up a performance scarcely better in the ‘Pentekontaetia’ than Herodotus had done for the seventh and sixth centuries B.C. (s. 89–117) in spite of his criticisms of his great predecessor (1. 20. 3). The new interest in dates as such seems to have arisen in the latter part of the fifth century B.C., with such historians as Hellanicus and the compiler of Olympic lists, Hippias of Elis. It is instructive that that good ancient scholar Plutarch was very sceptical of the early items in these lists (Plutarch, Numa),. The archon dates in the A.P. are the product of the same tendency among the Atthidographers. Cf. F. Jacoby, Atthis 1949, Passim; Strasburger, H., ‘Herodots Zeitrechnung’ Histosia, v (1956), 129–61;Google ScholarMiller, M., Klio xliii (1965), 109–25;Google ScholarBoer, W. den, Mnemosyne, xx (1967), 30–60.CrossRefGoogle Scholar

^{page 14 note 1} Gomme, , J.H.S. xlvi (1926), 173–8.Google Scholar

^{page 16 note 1} M. Lang, loc. cit. (above, p. 2, n. 2), 66–7; Miller and Sumner, locc. citt. The point here is that according to Herodotus (1. 61), Hippias and Hipparchus were ‘youths’ old enough to object to the father's marriage with Megacles' daughter. Therefore they should have been born fifteen to twenty years before that event. See now J. K. Davies, Ath. Prop. Fam. 445–8.

^{page 16 note 2} Davies distinguishes Thettalos from Hegesistratus, following Thucydides' statement (5. 55. 1) that the former was a brother of Hippias and Hipparchus, while Hegesistratus was the non-citizen son by an Argive wife (Hdt. 5. 94. i). He believes that A.P. (17. 3–4 and 18. 2) has conflated the two (Ath. Prop. Fam. 448–9).

^{page 17 note 1} Gomme, loc. cit. 174, and note 7. Lucian in his treatise Makrobioi has an interesting section (10–17) on princes who lived to an active old age. Among these were Ateas and Hieron of Syracuse. Hippias himself does not figure in the fairly long list, probably because neither Herodotus nor A.P. gives his exact age when he died. He was, however, (Hdt. 6. 107, supported by Thuc. 6. 59. 4) and was the son of a tough father who also lived to a ripe old age (A.P. 17. 1).