K. H. Kinzl home page TLG code for Greek text
AP 22.4: The Sole Source of
Klio: Beiträge zur Alten Geschichte. Herausgegeben von der Akademie der Wissenschaften der DDR, Zentralinstitut für Alte Geschichte und Archäologie. Berlin: Akademie-Verlag. Heft 1, 1991, Band 73
© Akademie-Verlag, Berlin
1. Ostrakismos is a well-known fact in Athenian history on which scores of studies have been written. Ostraka have been recovered. They can be interpreted in only one way, which is that suggested by the literary source:2 votes, cast by Athenians, in a procedure that sent citizens into a peculiar kind of exile. The literary tradition is thus vindicated by the archaeological evidence, as regards the voting with ostraka. The chronological limits suggested by the literary tradition are also confirmed. All ostraka date from the fifth century.3 The evidence of these ballots also provides the names of all but one (surely spurious4) of those identified by the literary sources.
That, however, is the whole extent to which the physical evidence of the inscribed potsherds conforms with the literary evidence of our sources and is suited to comple-
ment the latter. In all other respects we are at the mercy of the literary sources, which range from Herodotos to the lexica and other productions of late antiquity and the Byzantine age. With regard to nearly all of the most important questions asked by the historian (e.g., when was the law on ostrakismos introduced, for what reason and purpose, on whose initiative), it is only by recourse to critical evaluation of the literary evidence, and secondly by historical argument based on it, that we may attempt to suggest answers.
It is problems in the first category, that of a critical evaluation of the sources, with which I am concerned in this paper. I have become convinced that the force of a historical argument is not enhanced by merely citing a literary source. First and foremost, the intrinsic validity of the literary source as historical evidence must be objectively assessed. Whatever we know, or strive to know, as historians, must be suppressed: it must not interfere with the process of establishing whether the literary source before us is sound.5
The antiquity of a literary text (which, in many instances, goes back no farther than the Byzantine age) does not absolve us of the duty to subject [it to] a ferocious internal criticism, even if the result of that process is that nothing then remains.6
The present investigation focusses on two passages:
(a)Harpokration s. v. *I(/PPARXOS, and especially the section dealing with Hipparkhos son of Kharmos (which appears in FGrHist as Androtion 324 F 6);
2. The lexicon on the Ten Orators by Harpokration is extant in two forms: an epitoma, and a full version which, however, is both contaminated and abridged.7 I shall refer to the extant full version throughout this study by the name Harpokration, in order to remind the reader that no clean copy of Harpokration is extant. The gloss s. v. *I(/PPARXOS appears in both versions; it is only in the full version, however, that the reference to Lykourgos and Androtion in the section on which our investigation focusses are preserved.
The complete gloss of Harpokration s. v.*I(/PPARXOS reads as follows:8
[A] *DHMOSQE/NHS E)N TW=I KATA\ *KTHSIFW=NTOS. TU/RANNOS H)=N OU(=TOS *E)RETRI/AS.
This text displays some characteristics of learned explanation of Attic O)NO/MATA. It is assumed that Harpokration depends on a source of the first order, compiled from authoritative sources: an Attic onomasticon,9 commonly known as V1. Its author has recently been identified by Alpers,9a on circumstantial evidence, as Ioulianos, a near contemporary of Oualerios Diodoros (who can be dated by his purchase of a boat, on 173 03 31, in Oxyrhynkhos10). Our Oualerios Harpokration belongs to the Verwandtschaftskreis of this Diodoros.11 V1 also lurks in the onomasticon by Polydeukes (= Pollux); the very late Lexicon rhetoricum Cantabrigiense11a (which is extant in the margins of a Cambridge manuscript of Harpokration); Lexicon Patmense;11b POxy 1804 (probably Diodoros12); and, as the source for factual information, explaining sachliche Schwierigkeiten in the orators, in the onomastic glosses of *LE/CEIS R(HTORIKAI/. For this prime source of the fifth lexicon in Bekkers Anecdota, Wentzel invented the symbol V1,13 whose practice I shall follow.
Does the entry *I(/PPARXOS in Harpokration deserve to be praised as excellent and extremely learned (attributes that attach to V1)?
In the concluding section [E] of our gloss, Harpokration adds, to his explanations on four individuals with the name *I(/PPARXOS, a comment on the Athenian office of the I(/PPARXOS. Under the lemma I(/PPARXOS, most lexica explain only the office.14 Polydeukes 8.94 and Photios s. v. I(/PPARXOI deal also, in the same gloss, with the offices of the FU/LARXOS and the TACI/ARXOS14a, preserving V1s grouping together of factual information on related subjects. Only Harpokration joins the explanation of the military office to another group of glosses (Sammelglosse), in which information on persons named *I(/PPARXOS is provided; and only Harpokration preserves references to sources. The corresponding commentary on the military officer styled I(/PPARXOS in the Lexicon Patmense, and that of a Demosthenes scholion, are attached to Demosth.21.164, where the explanation of the term is useful indeed.14b Harpokration, however, refers to Demosthenes First Philippic (4.26). This reference appears inapposite, because the factual information provided by Harpokration is no more detailed than that already given by Demosthenes at 4.26.14c Harpokrations reference to AP is equally gratuitous. APs facts add nothing to Demosthenes remarks, and there is nothing in Harpokration that could be derived only from AP (61.4).14d
Under his separate lemma FU/LARXOS, Harpokration cites the same sources as in the sub-gloss s. v. *I(/PPARXOS (*DHMOSQE.NHS *FILIPPIKOI=S, preceding his explanation regarding FU/LARXOS, and, following it, W(S *A)RISTOTE/LHS E)N TH=I *A)QHNAI/WN POLITEI/AI FHSI/). No other extant gloss s. v. *FU/LARXOS contains any references to sources (except, copying the Harpokration epitoma with its reference to only Demosthenes Philippics, Souda (*F 82915). Harpokrations explanation of FU/LARXOS is virtually identical to that in the *ETUMOLOGIKO\N *ME/GA, and to the one in *LE/CEIS R(HTORIKAI/. This group of glosses is significantly different from Polydeukes 8.94 and Photios s.v. I(/PPARXOI whose text is nearly identical with AP 61.5 (but who do not cite AP). The reference to AP in Harpokration s.v. FU/LARXOS is again (as that s.v. *I(/PPARXOS, [E] inappropriate. Harpokration contains the information that the FU/LARXOS is U(POTETAGME/NOS TW=I I(PPA/RXWI (which also occurs in *LE/CEIS R(HTORIKAI/, the Souda, and EM, s.v.). AP 61.5, however, does not contain such a statement.16
It appears that there existed two Sammelglossen. One offered extensive commentary on Demosth. 4.26, drawing, at least partly, on AP, and with the appropriate references. The other provided information on various individuals of the name *I(/PPARXOS encountered in the speeches of the Ten Orators. At least the former of these Sammelglossen most likely made its debut in a commentary on Demosthenes by Didymos. The latter, more likely, first appeared in a work on the Ten Orators. Remnants of the summary gloss which provided commentary on Demosth. 4.26 are, stripped of the references to the passage where the words occur and to the sources of the information, preserved by Polydeukes and Photios; also, at a greater distance from Didymos, in the form of glosses under separate lemmata (s. vv. I(/PPARXOS, FU/LARXOS, and TACI/ARXOS, or at least under one or two of them), in other lexicographical sources, as seen above. In the case of the Lexicon Patmense, which preserves only the commentary on I(/PPARXOS, the explanation is sensibly transferred to Demosth. 21.164. Harpokration, however, who also tears the I(/PPARXOS portion of the Didymos commentary on Demosth. 4.26 from its original context, retains the reference to Demosth. 4.26. Harpokration furthermore reattaches this fragment (with the reference) to the other Sammelglosse on various individuals of the name *I(/PPARXOS (which may have been transmitted to him through different channels). The reference to AP, however, is retained by Harpokration, from the original summary gloss on Athenian military positions. It is, at its new location, no longer pertinent, once the Didymean Sammel- gloss had been mutilated and its substance contaminated.17
The sum of this exploration amounts to a note of caution. It would appear that Harpokration cannot be classified as a source of the best order, and one in whose information we are fully justified in placing great confidence.17a
In the section which is central to our investigation ([*G]), Harpokration comments on Hipparkhos son of Kharmos, and cites Lykourgos and Androtion. Our confidence in the text and the testimony of Harpokration is weakened by a simple line-by-line
comparison of Harpokration and the Souda. The gloss of the Souda is amongst those for which the author of the Souda uses Harpokration as his source. It is a Harpokration epitoma, however, which provides the Harpokration glosses in the Souda.18 That epitoma, dated to the ninth century, is based on a manuscript of Harpokration superior to those which preserve our extant Harpokration.19 It is illuminating to contemplate the range of stages at which a particular infelicity in the Souda might have been produced. It may be owed to V1s treatment of his source; or to Harpokrations treatment of V1, or to Harpokration s treatment of Harpokration, or to the epitomists treatment of Harpokration, or to Soudas treatment of the epitoma. (In the following comparison of Harpokration and the Souda, I underline phrases in Harpokration which are absent from the Souda).
Harpokrations ME/N after SUGGNH/S has dropped from epitoma/Souda. It is without a corresponding DE/ in Harpokration (even though it might appear as if the subsequent KAI/ serves this purpose). The epitomist apparently judged the ME/N superfluous, if not inappropriate. The dangling ME/N seems to betray a change of sources in the pre- Harpokration tradition, from a source which explained the genealogy of the first ostracisé Hipparkhos son of Kharmos, to another source which explained his ostrakismos and why ostrakismos was introduced.20 The result is an example of poor writing rendered poorer still by the tacking on of a causal preposition, to which a causal clause is attached.
Harpokrations DIA/ was thought by the epitomist to go with E)CWSTRAKI/SQH, rather than with TEQE/NTOS. The epitomist thought therefore that he could safely eliminate the entire phrase TOU= ... NO/MOU ... TEQE/NTOS of Harpokration. It is only because we possess the text of AP 22.320a that we are able to determine that the epitomist completely distorts the meaning of the text he epitomises. It is possible furthermore that the epitomist was uncomfortable at the thought of a first introduction of a law,21 for which reason he deleted the entire phrase more readily.
Harpokrations TW=N PERI\ *PEISI/STRATON also fell victim to our epitomists sense of what he thought he was reading in Harpokration. We are here provided with an excellent example of the way in which men of some learning approached their texts. In strictly logical terms, the alteration is an improvement. The gloss speaks of Hipparkhos family ties with Peisistratos, and of Peisistratos ascent to the tyranny as a result of his rôle as a DHMAGWGO\S KAI\ STRATHGO/S, but of suspicion not of Peisistratos but of TW=N PERI\ *PEISI/STRATON. Our learned epitomist, disturbed by this seeming inconsistency without hesitation hastens to correct it. It is only the preservation of Harpokrations gloss, of AP, and generally our historical knowledge built up from a variety of sources, that saves us from taking the lines in the Harpokration epitoma and the Souda as a statement of historical fact. There are of course innumerable cases in which a late lexicographical gloss is our one and only source of information. In such cases the historian who inclines to the fundamentalist position accepts the late lexicographical sources testimony as a statement of fact. The lexicographer is not, however, a historian, neither in the ancient tradition, nor, above all, in the modern sense of the word. He is a mere medium, the historical value of whose message must be proved.
Attempts at mending passages in lexicographical sources are frequently misguided.22 Neither our sense of good Greek, nor an appeal to our knowledge of the real course of events, justify interfering with a text that is demonstrably flawed but which is flawed because its author wrote a flawed text. Lexicographical sources are, almost as a rule, riddled with errors of fact and poor prose.23
It is the genesis of a text which is the only valid subject of investigation. Whether it means this or that, or is written in good Greek, let alone whether it makes any historical sense, these are irrelevant questions. The only valid question is that after the lines of transmission of the text, the source to which they. ultimately lead us back, and the degree of reliability in the preservation of the original source. The lexico-
grapher (neither an eyewitness to events, nor a historian, but a grammarian with the specific interests of the grammarian) merely passes on to us a certain text containing certain information in a certain state of preservation.
In the case of our Harpokration gloss, both internal analysis of the gloss and comparison with the Souda gloss have produced adequate grounds to suggest that the gloss must be approached with extreme caution. Wentzels characterisation of V1 as an excellent and extremely learned onomasticon cannot be, as a matter of course, translated also to apply to Harpokrations gloss s.v. *I(/PPARXOS.
3. The section of the Harpokration gloss which deals with Hipparkhos son of Kharmos purports to elucidate Lyk., In Leocr. 117. There, Lykourgos edifies his audience by rehearsing how the ancestors waged war on traitors, citing specifically the treatment of Hipparkhos son of Kharmos. Hipparkhos was accused of treason and, when he failed personally to face the charges, condemned to death in absentia. His statue (clearly not a statue of Hipparkhos but one dedicated by Hipparkhos if there is any historical truth to the entire story) was taken down from the Akropolis, melted down, and turned into a bronze stele. It was decreed to inscribe on this stele the names of all A)LITH/RIOI and PRODO/TAI, including the name of Hipparkhos son of Kharmos. This is not a suitable context in which to mention Hipparkhos ostrakismos. It is ostrakismos, however, which is introduced in Harpokrations commentary who otherwise fails to throw light on any of those points on which elucidation might be welcomed.24 Mention of Hipparkhos ostrakismos in Harpokration is as misplaced as it would be in the speech of Lykourgos. If a gloss whose source is an onomasticon of the excellent quality of V1 a mentions only Hipparkhos family bonds with Peisistratos and his ostrakismos, we may infer (unless we assume that Harpokration reproduces only the most irrelevant portions of V1s commentary) that this is all the information which V1 was able to gather in his sources,25 and which he duly passes along to his readers. Had V1 been able to gather other, more directly relevant, information, he presumably would not have mentioned ostrakismos, and especially not the reasons for its introduction, in the first instance. We need not be surprised by this. Hipparkhos principal distinction was his name.26 If there was any debate about him in the now lost sources, that debate would most likely have turned on his name, Hipparkhos.
An onomasticon of the quality of V127 might be expected to elucidate the name, and therefore the genealogy, of Hipparkhos son of Kharmos. Since Lykourgos himself contains no information, V1 had to draw on some other source or sources.
Kleidemos informs us that Hipparkhos father was the lover of Hippias, and that Peisistratos had arranged for Kharmos daughter to marry his son Hippias. According to Ploutarkhos, Peisistratos was the lover of Kharmos. Pausanias informs us that Kharmos dedicated an altar with an accompanying epigram, which is also quoted by Kleidemos (who, however, speaks of the dedication of a statue of Eros).28 In Lykourgos,29
there is nothing at all on the background of Hipparkhos and his father Kharmos, be it familial, erotic, or otherwise.
Thoukydides much analysed digression on the Peisistratidai30 proves that both the finer points of the genealogy and the erotic exploits of the Peisistratidai were extensively discussed.
AP 22.6, in summing up the ostrakismoi of the first three victims, refers to them as TW=N TURA/NNWN FI/LOI, as he does at 22.4 where he marvels at the lenience of the people, who permitted them to remain in Athens. Despite this summary characterisation of the first three ostracises as TW=N TURA/NNWN FI/LOI, AP 22.4 identifies Hipparkhos son of Kharmos not as a TW=N TURA/NNWN FI/LOS but as TW=N E)KEI/NOU [viz. *PEISISTRA/TOU] SUGGENW=N. This contradiction might indicate that the knowledge of Hipparkhos family ties to Peisistratos represents a more recent accretion to the received common knowledge about Hipparkhos and the Peisistratidai.
The date for this enlargement of historical knowledge falls between the dates of composition of Kleidemos Atthis and AP.31 The most likely candidate, thanks to Harpokrations citation, is Androtion.
We may assume that Androtion is the discoverer of the bonds of kinship between Hipparkhos and Peisistratos. In a work of the compass of his Atthis,32 and in view of the extensive treatment of the subject by Thoukydides,33 he expounded his discovery at considerable length, and in a suitable context.34 It would seem most likely that Androtion dealt with Peisistratid genealogy, which included Hipparkhos son of Kharmos, in a Peisistratid context, i.e. no later than the expulsion of the Peisistratidai from Athens.
Ostrakismos and the reasons for its introduction were treated by Androtion also at some length. The most suitable context for its treatment was either under the date of its introduction, or at the point at which it was first applied, leading to the ostrakismos of Hipparkhos.
Androtions discourse on Peisistratid kinship connexions and that on the origins
of ostrakismos would thus appear to have occurred at widely separate places in his work, and at much greater length than AP 22.3ff. AP 22.3ff. therefore cannot, in one piece and at equal length, be derived from Androtion. The text of AP is, in its wording, too far removed from its source or sources to permit a positive identification. It is not unlikely, however, that APs characterisation of Hipparkhos as a SUGGENH/S of Peisistratos is somehow owed to Androtion.35
For inclusion in a gloss on Hipparkhos, in a lexicon on the Ten Orators, Androtions discourse on Peisistratid genealogy is unsuited. If Thoukydides digression is any guide, even an extremely learned lexicon such as that of V1 could not accommodate Androtions lengthy text.
The reference to Androtion in Harpokration is, however, an undeniable fact. How can we account for it?
It is true that there are references in lexicographical sources which are inapposite, inaccurate, or even plain wrong.36 This may be the result of a variety of causes, such as errors that occur in the process of the transmission of a text or of abridging a source. It is contrary to the very purpose of a lexicographical gloss, however, intentionally to give a false reference. Harpokration surely does not attempt to pull the wool over our eyes regarding his true source. If Harpokration refers to Androtion, and if the reference is not caused by a textual corruption, Androtion must have contained information on some aspect of the gloss. It seems unlikely that Androtions name is to be connected with Harpokrations explanation of the origins of ostrakismos, however. Our analysis above suggests [considerable] problems with the text of Harpokrations gloss, especially a possible change of sources after SUGGENH\S ME/N. We also attempted to demonstrate that Androtion is unlikely to have dealt with Peisistratid genealogy and the origins of ostrakismos in one brief passage. It is only Harpokrations statement of the fact that Hipparkhos son of Kharmos was a SUGGENH/S of Peisistratos, to which the reference to Androtion applies. The reference does not imply that Harpokration quotes the words of Androtion. The reference also does not extend to Hipparkhos ostrakismos and the conjoined explanation of the reasons for the introduction of the law on ostrakismos.36a With these qualifications, the reference to Androtion in Harpokration may be regarded as in order.
4. A close resemblance between Harpokration and AP was recognised almost instantly after the publication of PLond 131,36b and attention was drawn to it again, more recently, by several scholars.37 It is only by means of a line-by-line comparison of the two texts, however, that it becomes evident that they are made up of virtually the same words. Of APs text of thirty-five words and Harpokrations text of (not counting the introduction to the sub-gloss and the reference to Lykourgos and Androtion) thirty-two words, no fewer than twenty-six are shared between them. The only
difference that might be worth mentioning consists of changes in the sequence in which these shared words occur.
I shall demonstrate the truly amazing near-identity of the two texts at some length, because this has never before been attempted, not even by those scholars who seem to have perceived it broadly.38 By its mere lengthiness, this procedure ought to succeed in proving beyond doubt the true nature of the relationship between AP and Harpokration. Harpokration is derived from no other source than AP.39
First, I shall print APs original continuous text, with the corresponding segments of Harpokration in the lines below it (vertical strokes with small numerals separate these chunks from one another, and indicate their original sequence in Harpokrations text; words in Harpokration which occur outside these chunks are, in addition, inserted in parenthesis beneath the corresponding words in AP). Second, I shall reverse the procedure, by printing Harpokrations original continuous text, with the corresponding segments of AP in the lines below (and in the same way as in the first case).
Harpokrations testimony regarding Androtion lacks authority, regarding both substance and actual wording.40 It is noteworthy that Jacoby prints Harpokrations
text not (as is his practice with texts which he regards as true quotations) gesperrt: Jacoby clearly suggests that Harpokration preserves not the actual words but only the substance of Androtion, and he inadvertently produces a compelling argument against common Androtionian origin for Harpokration and AP.
Since the two texts are not completely identical, one or the other must have undergone alteration of some kind. It is almost unthinkable that, of the two texts, Harpokration would be the one which preserves Androtion unaltered.40a Jacoby evidently thought otherwise. AP, however, is even less likely to have copied Androtion verbatim.41 If neither Harpokration nor AP reproduce a copy of Androtions text, and since the orthodox view also excludes the possibility that Harpokration copied from AP, how can it be explained that Harpokration and AP, preserving the substance, but not the words, of their alleged common source, Androtion, produce at the same time texts in which miraculously some seventy-five to eight-five percent of the words they use are identical?
It is only by abandoning the orthodox position42 and acknowledging the obvious, that the inexplicable becomes comprehensible AP, not Androtion, is the source of which Harpokration represents an adapted version.
I would explain the flawed Harpokration gloss either as an intentionally (and incompetently) abridged or altered version, or as a simple corruption, or both, of the original gloss of either Harpokration or, more probable, V1. That original gloss was closer to, if not identical (so far as identical goes with lexicographers) with, AP.
There is in any event no good reason to continue to believe that Harpokration reproduces a text written by Androtion.43 Androtion cannot have written anything resembling the poor prose which constitutes the passage in Harpokration.43a There is accordingly no ground on which to judge Harpokration a source that permits us to make assumptions about the opinion of Androtion. Once our confidence in Harpokrations reliability is shaken, however, we must dismiss his text as a basis for serious discussion. We cannot treat Harpokration as a smorgas board, to which we may help ourselves, selecting what suits our tastes but leaving behind that which does not. Harpokration is a source unfit to vouch for the opinion of Androtion on the date of the passage of the law on ostrakismos.
5. Let us now direct our attention to the passage of AP, which is allegedly derived from Androtion, because Harpokration allegedly provides incontrovertible proof.
AP 22.3-6, on close inspection, reveals qualities of literary composition worthy of a good writer. It is coherent, lucid, and not flawed in its logic (it is irrelevant to the argument, whether, as historians, we agree with the account43b). Its composition is such that it cannot be dismissed as a patch-work of carelessly cobbled together elements, borrowed piecemeal, from this or that source. There is evidence of carefully thought out composition, including Ringkomposition.44 It proves that the entire passage is of one mold, written by one author alone. This will be shown in the analysis of AP 22.3-6 reproduced on p. 41.45
6. My lengthy demonstration will be received with the argument that I only prove an already well-established fact: the evident proximity of the texts of AP and Harpokration is the obvious consequence of their sharing one and the same source, i.e. Androtion. Although several scholars have attempted to devalue the testimony of the fragment of Androtion in Harpokration, the weight of their argument is lessened because they insist on deriving Harpokration from Androtion, and because they persist in mingling the arguments of the historian with those of the Quellenkritiker.46
According to the orthodox creed, the close resemblance between AP and Harpokration must be viewed as the consequence of a derivation of both texts from their common source Androtion.47 By embracing the orthodox creed, scholars have prevented themselves from contemplating alternatives, especially derivation of Harpokration from AP, rather than from Androtion.48 It has also effectively precluded further debate on any question connected with the entire complex of both quellen-
kritische and historical problems, unless that debate incorporates as its premiss the orthodox view.
The onus of proof, however, is on those who adhere to the orthodox creed. They ought convincingly to reply to the following questions.
(a) Is it the practice of the author of AP to reproduce his sources in the same way as grammarians, scholiasts, lexicographers, etc.?49
(b) Is the following procedure likely, or even conceivable: Aristotle dates the introduction50 of the law on ostrakismos to Kleisthenes. In his explanation of the reasons of its introduction he uses a source that dates it to the time of the ostrakismos of Hipparkhos. Aristotle copies his source verbatim, but transposes the two little words TO/TE PRW=TON. By this device, he alters the substance of his source on a fundamentally important and potentially controversial issue. He refrains from indicating his disagreement with his source, and from putting forward any argument against it.
(c) If there was a controversy on the date between Aristotle and the author whose text he copied, how are we to explain it that it has left no trace in our numerous sources?
Let me put forward my own answers to these questions.
(a) There are instances in which we are still able to compare the (direct or indirect) source of AP with the actual text of AP. These examples show very clearly that the author of AP, even when he stays very close to the original, does not resort to verbatim copying, changing, at his most daring, the word order.51
(b)Aristotles position on ostrakismos is ambivalent. At pol. 1284 b and in AP 22.3, he is far from offering an outright condemnation of the law. He expresses disapproval, however, at AP 22. 1, citing the law on ostrakismos specifically as evidence for Kleisthenes catering to the whim of the masses. The words by which he introduces the first series of ostrakismoi also cannot be construed as expressing his approval of the procedure (QARROU=NTOS H)/DH TOU= DH/MOU, 22.3). On the other hand, he draws a clear distinction between the anti- tyrannical sentiment which led to the introduction of the law and its actual use in the first three ostrakismoi (22.3-6), and the beginnings of the radical democratic abuse with the ostrakismoi of those A)/PWQEN TH=W TURANNI/DOS (22.6). It is the latter practice which Aristotle anticipates at 22.1, and on the evidence of which he characterises Kleisthenes as STOXAZO/MENOS TOU= PLH/QOUS.
Aristotle would have been able to construct a more gradual progression of from Solon, to a more DHMOTIKO/S Kleisthenes who, however, would still be a long distance away from the objectionable Ephialtes and his move to demagogy. Ignoring his own qualification at 22.3, he judges ostrakismos as radical democratic and
consequently places Kleisthenes much nearer Ephialtes than he might otherwise have been inclined.
Ostrakismos plays a crucial rôele in Aristotles portrayal of Kleisthenes democracy as radical democratic. Ostrakismos is not merely one element amongst several, cited only to serve as an illustration, and without which Kleisthenes would still rank as the founder of radical democracy. Aristotle did not judge Kleisthenes democracy on a priori grounds as radical. It is ostrakismos which is radical, and whose father must therefore be termed radical.
Aristotles line of argument is laboured and inconsistent. A consistently gradual progression from Solon to Kleisthenes to Ephialtes could have been put forward, had it been possible to disconnect ostrakismos from Kleisthenes. Had Aristotle been able to date the introduction of ostrakismos to the post-Kleisthenic period on good authority such as Androtions, he probably would have followed that lead. This would have allowed him to produce a less laboured and inconsistent account. The less than straightforward line of argument in Aristotle appears caused by not incompetence but the fact that ostrakismos was inseparably linked to Kleisthenes.52
My hypothesis regarding Aristotles assessment of the political significance of the democracy of Kleisthenes in AP may be rejected by others, and perhaps even proved wrong. It would none the less remain an utter mystery and improbable in the extreme, had Aristotle chosen the procedure which is implicitly postulated by those who believe that his account at 22.3- 4 was copied from Androtion. That hypothesis requires us to believe not only that Aristotle was in disagreement with Androtion on an important date in Athenian constitutional history (entirely conceivable in itself) but also that Aristotle rejected Androtions date without producing any evidence or stating any reasons for his rejection; and that he never the less copied that account with which he disagreed, with the exception of the two little words TO/TE PRW=TON which he merely shifted around in that text in order to make it mean the
opposite of what its original author meant. Had Aristotle disagreed with Androtion but felt disinclined to engage in an argument against him, or indeed to compose his own text, he would, by any standard of normal human behaviour, have selected an account for copying with which he could agree on all counts.
(c) There is no indication in any of our numerous53 sources for ostrakismos that the date of its introduction was not universally agreed upon. Had there been disagreement, there would have been debate.53a The texts which deal with ostrakismos are, almost without exception, of the antiquarian kind. To the antiquarian mind questions such as that after the date of the introduction of the law on ostrakismos are of intrinsic significance and demand attention.54. Apart from Harpokration with the fragment from Androtion, which cannot be adduced as evidence55 for a disagreement on that date, there is no source that disconnects ostrakismos from Kleisthenes. There is accordingly also no source that hints at a debate on the date.
To suggest that there was disagreement on the date between Androtion and AP and that it (and the discussion on it) has vanished without a trace, stretches the imagination of anyone familiar with the lexicographical sources.55a Harpokration cites both Androtion and AP in our gloss. It is most improbable that his learned source excerpted information from both authors but passed up the opportunity of demonstrating his learning by pointing to the controversy on the date (or, even more improbable, failed to notice the divergence of opinion) between Androtion and AP; or, if he did, that Harpokration cleanly excised the controversy.
The onus of proof is on those who insist that there was a difference of opinion. They must prove that Androtion put forward a non-Kleisthenic date; and, why this discrepancy was not duly noted; or, if it was, why no trace of the notice survives. The argumentum e silentio is, in this instance, a powerful one; it is not of the more specious kind.
7. Let me sum up the results of my examination of AP 22.3f. and Harpokration s.v. *I(/PPARXOS (with Androtion, FGrHist 324 F 6), passages central to any argument of the historian in establishing his own date for the introduction of the law on ostrakismos which is one either in the context of the Kleisthenic reforms, or immediately preceding the first ostrakismos (i.e., that of Hipparkhos son of Kharmos, in 488/487). The historian ought to take the following conclusions into account.
(a) Harpokration, whilst entirely correctly citing Androtion as an authority, cites
Androtion as an authority only on the kinship bonds between Hipparkhos son of Kharmos and Peisistratos. It is only these kinship bonds regarding which we may be confident that the name of Androtion vouches, be it in Harpokration or in AP.
(b) The text of Harpokrations gloss on Hipparkhos son of Kharmos (elucidating his family background, his ostrakismos, and the reasons for the introduction of the law on ostrakismos) is in its entirety derived from AP 22.3f., representing a near-verbatim, but flawed, rendering of APs text.
(c) No source is extant that documents positively the opinion of Androtion on the date of the introduction of the law on ostrakismos.
(d) In the absence of any positive evidence to the contrary, there is every reason to conclude that Androtion not only did not disagree but indeed was in complete agreement with every other extant source for the date of the introduction of the law on ostrakismos. Androtion too believed in the Kleisthenic origin, as part and parcel of the reforms of Kleisthenes, of the law on ostrakismos.
Although I argued against attempts to emend Harpokration,56 the results of my investigation demand one, and strongly suggest another, emendation of Harpokrations text.
First, the punctuation between the remarks on the family relationship and the ostrakismos of Hipparkhos must be changed, to a full stop: ...*A)NDROTI/WN ... FHSI\N O(/TI SUGGENH\S ME\N H)=N *PEISISTRA/TOU TOU= TURA/NNOU. KAI\ PRW=TOS E)CWSTRAKI/SQH ...
Second, the words at the end of the gloss (W(/S FHSIN *A)RISTOTE/LHS E)N *A)QHNAI/WN POLITEI/AI, referring to the Athenian office of the I(/PPARXOS) appear misplaced. AP does not state what Harpokration suggests he does (whilst the other lexicographical sources do not cite any source).57
I would suggest that the reference to AP appears in that particular place only because of some corruption. It was moved there from its original place at the end of the sub-gloss on Hipparkhos son of Kharmos. A scribe may have omitted the reference and then tacked it on at the end to cover up this error. Or a scholar who corrected a scribes work who had omitted the reference inserted it in the wrong place. Or a scholar conjectured that the right place for a reference could only be the sub-gloss on the Athenian office, and emended the text by moving the reference to its final resting place at the very end of the gloss, after the reference to the Athenian office.
I would therefore suggest to transplant the words W(S ... POLITEI/AI from the sub-gloss on the I(/PPARXOS to the end of the sub-gloss on Hipparkhos son of Kharmos.58
I would consequently suggest to emend the sub-gloss on Hipparkhos son of Kharmos to read as follows:
1 This paper results from a complete re-examination of the sources on ostrakismos, last undertaken admirably (although I cannot agree with his identification of the ultimate source of all we know as Theophrastos) by A.E. Raubitschek, Theophrastos on ostracism C&M 19 (1958) 73-109. Because of the considerably size of the investigation, I decided to publish the portion on Hipparkhos separately. As I am concerned exclusively with problems of texts, their sources and transmission but not with historical problems, references to the innumerable articles on ostrakismos as a historical problem (dealing with the sources only in the context of historical enquiry) have been omitted. Gratitude is due to William Slater who generously gave of his time by reading and commenting on this paper; his agreement with any part of it, however, must not be assumed. I am also grateful to my colleague Ian Storey for reading an earlier draft.
2 The list of primary sources on ostrakismos is too long to be here repeated. Virtually everything of any significance is in Raubitschek (n. 1). The so-called new evidence of Vat. Gr. 1144 foll. 222r-222v, no 213 of an Appendix Vaticana, first published by L. Sternbach, [at the present time HTML is unfortunately not capable of representing all Polish characters correctly; for the correct version see the print edition] Gnomologium Parisininum ineditum Rozprawy akademii umiejetnosci: wydzial filologiczny (Krakow: Akademia umiejetnosci) ser. 2, tom 5, 1894, 135-218, at 192, is fortunately of no concern to us here. I hope to be able soon to argue (cf. n. 1) that the text is a worthless distortion (apart from being corrupt) of information drawn from the one single extant tradition on ostrakismos which surfaces in a variety of shapes in all extant sources dealing with ostrakismos (the thesis of Raubitschek [ n. 1] must be upheld in this respect). For some salutary scepticism in the face of surprising enthusiasm following its republication by J. J. Keaney and A. E. Raubitschek, A late Byzantine account of ostracism AJPh 93 (1972) 87-91 (with in several ways significantly inaccurate bibliographic information, corrected above, at 87 n. 3), see P. J. Rhodes, Comm. AP 268; also, R. Develin, Cleisthenes and ostracism Antichthon 11 (1977) 10-21.
3 The most recent Tabulation of the sherd evidence is still R. Thomsen, The origin of ostracism, København 1972 (Humanitas 4), 71-80.
4 *MILTIA/DHS O( *KI/MWNOS, Andok. 3.3.
5 My work on problems in the sources for historical events has persuaded me of the necessity to treat the sources in isolation from historical considerations. See Miltiades-Forschungen, Wien 1968; Philochoros FGrHist 328 F 115 and Ephoros Hermes 102 (1974) 170-190 (esp. 184); Miltiades Parosexpedition in der Geschichtsschreibung Hermes 104 (1976) 280- 307. My position as a historian regarding the date of the introduction of the law on ostrakismos has no bearing on the present investigation (Athens: between tyranny and democracy Greece and the eastern Mediterranean, Berlin - New York 1977, 209).
6 The quotations are from J. N. Coldstream, The formation of the Greek polis: Aristotle and archaeology, Opladen 1984 (Rhein.-Westf„l. Ak. d. Wiss.: Geisteswiss. Vortr„ge, G 272), 7 (protesting such criticism). See esp. M. I. Finley, The ancient historian and his sources id., Ancient history: evidence and models, London 1985, 7-26. Cf. my study On the consequences of following AP 21.3 (on the phylai of Attika) Chiron 19 (1989), 347-365; also my review of Rhodes (Comm. AP) Gymnasium 92 (1985), 220; and my review article of W. Gawantka, Die sogenannte Polis, Stuttgart 1985 EMC n.s. 7 (1988) 409f., on Coldstream op. cit.
7 H. Schultz RE 7 (Stuttgart 1912), 2412f. s. v. Harpokration 5. Also, recent and authoritative, K. Alpers, Das attizistische Lexikon des Oros, Berlin - New York 1981 (Sammlung griechischer und lateinischer Grammatiker 4), 116ff.
8 I. Bekker (ed.), Harpocration et Moeris, Berlin 1833. W. Dindorf (ed.), Harpocrationis Lexicon in decem oratores Atticos 1, Oxford 1853 (repr. Groningen 1969) is more easily accessible and more frequently quoted; it is also the only edition for the epitoma.
Line 1 Demosth. 18.295; cf. 9.58. 2-7 E(/TEROS E)TURA/NNEUSEN] = FGrHist 324 F 6 (cf. p. 38f.). 2 Lykourg. 10-11 F 6 Conomis (= F 67 Blass.). 4 Lykourg. In Leocr. 117f. (cf. p. 35). 4-7 See p. 32 f. (epitoma and Souda). 6 PRW=TON] see n. 21. 8 [Demosth.] 59.26. 11 Demosth. 4.26 (cf. p. 31f.). KAI\ *A)RISTOTE/LHS] AP 61.4 (cf. n. 14a; Appendix).
9 G.Wentzel, Beitr„ge zur Geschichte der griechischen Lexikographen SPAW 1895, 477-487 (repr. in: K. Latte and H. Erbse [eds], Lexica Graeca Minora [abbr. LexGrMin], Hildesheim 1965, 1-11), at 483 (= 7); Alpers (n. 7) 117.
11a In A. Nauck (ed.), Lexicon Vindobonense, St. Petersburg 1867 (repr. Hildesheim 1965), 329-358 (a P.P. Dobraeo editum); E.O. Houtsma (ed.), Lexicon rhetoricum Cantabrigiense, Diss. Leiden 1870 (= LexGrMin 61- 139). Four of the glosses in LexCantabr are also preserved, under the name Klaudios Kasilon, in E. Miller (ed.), Mélanges de littérature grecque ..., Paris 1868 (repr. Amsterdam 1965), 397f. LexGrMin 243f.).
11b Ed. I. Sakellion (= LexGrMin 140-165).
16 It might be worth the while of a qualified scholar to investigate the possibility that AP 61.3-7 in PLond 131 comes not from the original work of Aristotle but from a lexicographical source, i.e. Didymos, restoring in this way the text lost in a lacuna (but announced at 43.1: [X]EIROTONOU=SI DE\ KAI\ TA\S PRO\S [T]O\N PO/LEMON A(PA/SAS). The restorer was less fortunate with the elective offices whose term ran from Panathenaia to Panathenaia, l.c.: that lacuna (first posited by W.L. Newman CR 5 (1891) 163, and adopted by most editors) antedates 1the activities of Didymos and his likes (cf. Kaibel [n. 37] 250).
17 Even to attempt beginning to sort out the lexicographical tradition on these three terms (I(/PPARXOS, FU/LARXOS, TACI/ARXOS) is out of the question in this context. These glosses may, however, represent an interesting case, against which to test some established or controversial theories. E.g., the question whether Photios, like the Souda, exploited the Harpokration epitoma. Some scholars think he did, cf. A. Adler RE 4 A, 692f. s.v. Suidas 1; Schultz (n. 7) 2413; J.J. Keaney, The text of Androtion F 6 and the origin of ostracism Historia 19 (1970) 1-11, at 2 n. 5. Alpers (n. 7) 73 with n. 25 appears to sit on the fence (most of the material of Harpokrationian content comes from the erweiterte Fassung, whilst, vielleicht, and zum geringeren Teile, the Harpokration epitoma was used by Photios). C. Theodoridis, Photii Patriarchae Lexicon 1, Berlin - New York 1982, in his preface, does not list Harpokrations epitomised version amongst the sources of Photios.
It would appear prudent, however, not to place great trust in Harpokration as a reliable source for Androtions views, and not to regard Harpokrations references to Androtion and, especially, AP as cast in concrete. Cf., regarding Harpokrations references, p. 36f., concerning Androtion, and p. 45, concerning AP
21 E.g., G. V. Sumner, Androtion F 6 and Ath. Pol. 22 BICS 11 (1964) 82, citing H. Bloch Gnomon 31 (1959) 493 (review of F. Jacoby, FGrHist 3 b suppl.), who uses the term senseless. M. Chambers, Androtion F 6: TO/TE PRW=TON JHS 99 (1979) 151f., defending the traditionalist position on what Androtion allegedly says (below p. 38f.), attempts to show that so far as concerns Greek idiom, ..., TO/TE PRW=TON is neither meaningless or senseless, it is normal Greek (ibid. 152). At best, he proves that it is possible to read Harpokration in this way. He does not prove that it is impossible to interpret Harpokrations words as senseless: by quoting AP 22.3, Chambers invalidates his own assertion, because AP uses TO/TE PRW=TON in exactly that meaning which Chambers rejects for Harpokration.
Some have resorted to emending the text of Harpokration, guided by their belief that Harpokration preserves Androtion who cannot have written gibberish. Keaney (n. 17), referring the reader to id., Heliodorus F 1 and Philochorus F 41 Historia 17 (1968) 507-509, claims that TO/TE PRW=TON is merely a recent corruption or scribal emendation for TO/TE PRW/TOU that must have been the lectio of the last archetype (i.e., the manuscript from which all extant manuscripts and the Aldine edition are derived); TO/TE PRW/TOU, however, is meaningless and must be emended, op. cit. (n. 17) 3. Keaneys judgement appears a priori unsound because of his a priori assumption that Harpokration offers Androtions text. If that assumption is removed, the indubitably corrupt TO/TE PRW/TOU loses standing. If, furthermore, one assumes, as I believe one must, that Harpokration derives from AP TO/TE PRW/TOU is merely a corruption of TO/TE PRW=TON #&151; and there is no foundation for Keaneys elaborate emendation by which he restores a text of Androtion that was never written (NO/MOU PRO\ TOU/TOU TEQE/NTOS, l.c.; cf. Keaneys defence Androtion F 6 again Historia 25 (1976) 480-482; cf., incidentally, Chambers, ibid. 152 n. 8, although again arguing solely from possibility; there is an unfortunate misprint in l. 3 which renders his argument unintelligible: read PRW/TOU for PRW=TON).
23 The extant manuscript versions at any rate. I would not suggest that, for instance, the commentaries of Didymos were written in poor prose.
25 Dion. Hal. *A)RX. *R(WM. 5.77.6 and 6.1.1 mentions, without patronymic, Hipparkhos as the archon of 496/5 (cf. T.J. Cadoux JHS 68 (1948) 116). It would have been even more pointless to mention his archonship.
28 Athen. 13.609 C-D = Kleidemos FGrHist 323 F 15; Plout. Sol. 1.7; Pausan. 1.30.1.
29 See above.
30 Thouk. 1.53.3-60.1. The ganze Bibliothek (K. von Fritz, Griechische Geschichtsschreibung 1, Berlin 1967, 269 n. 51) written on this digression cannot here be referenced; I attempted to draw attention to the genealogical polemic, however, Zu Thukydides über die Peisistratidai Historia 22, (1973) 504-507.
34 F. Jacoby, Atthis, Oxford 1949 (repr. New York 1972), 337 n. 42 suggests that as Androtion calls this Hipparkhos SUGGENH\S *PEISISTRA/TOU he may have given particulars about the relationship. Attention may have been focussed on that point by Kleidemos who, however, did not contribute much to clarify matters (see above), whilst, on the other hand, preparing the ground for Androtions presumably more precise treatment of the question. It is one of those questions in Jacobys test case (the tradition on the Peisistratidai, ibid. ch. III § 2; § 4 [c]; at 152ff. and 188ff., respectively) regarding which Jacoby shows that later writers filled the gaps left in Herodotos work. Although I would be inclined to assume that Androtions new information is based on conjecture, others may think that he uncovered factual information hitherto unknown or ignored; my arguments in this paper are not about historical facts (see p. 29), nor about this particular problem, for which reason I shall refrain from arguing in favour or against the validity, as historical evidence, of Androtions claim that Hipparkhos was a SUGGENH/S of the Peisistratidai.
36b Cf. Rhodes, Comm. AP 3-5.
37 Esp. Sumner (n. 21) 79-86; K. J. Dover, Androtion on ostracism CR 13 (1963) 256f.; Thomsen (n. 3) 11-60 with extensive surveys of modern interpretations. Textual corruption. in Harpokration was observed by G. Kaibel, Stil und Text der *POLITEI/A *A)QHNAI/WN des Aristoteles, Berlin 1893, 174f.
39 The most explicit statement is Kaibels (n. 37) 175: ... deutlich ist, dass das scheinbare Androtionfragment nichts ist als ein elendes Excerpt aus Aristoteles [my italics]. For similar assessments by other scholars, see Thomsen (n. 3) 52.
40 See p. 31- 35. Further observations on points of detail only reinforce that judgement. For instance, Harpokration omits the demotic of Hipparkhos (he is admittedly not in the habit of giving demotics: a cursory cheek brings to light a total of four demotics, two of these belonging to ancient authors); even Ploutarkhos has a demotic (albeit a different and presumably the wrong one, at Nik. 11.8, a passage that must derive from a lexicographical source, as I hope to be able to demonstrate in a separate article, see above n. 1). The absence of Hipparkhos demotic would surprise in Androtions Atthis which, by the end of the second book, reached only the end of the first half of the fifth century (463/462 according to Jacoby, FGrHist 3 B suppl. text p. 105, 11. 2-6), and ran to a total of eight books (FGrHist 3 B suppl. text 104, 13, with suppl. notes p. 102 n. 136). At any rate, Androtions list of the ten generals at Samos in 441/440 names them with their demotics (schol. Aristeid. 3, 485, 30-3 Dindorf [at U(PER\R TW=N TETTA/RWN, vol. 3, 316, 7 Behr] = FGrHist 324 F 3).
41 See, e.g., the example cited at n. 51. The astute observation by Kaibel (n. 37) 175, that die an sich denkbare Möglichkeit, dass Arist.. den Androtion wörtlich ausgeschrieben habe, ist ausgeschlossen, is tempered, however, by Kaibels evident belief that AP 22.3f. none the less, in substance, and in its entirety, is derived from Androtion.
45 The brackets in the left- hand margin demonstrate the ring composition, the brackets in the right-hand margin tie recurring elements within the account together.
46 Cf. Thomsen (n. 3), and the literature cited there, esp. D. Kagan, The origin and purposes of ostracism Hesperia 30 (1961) 393-401 (cf. below n. 52). Kagans arguments in attempting to discredit the testimony of Harpokration are that: (a) the traditional interpretation of Harpokrations text, according to which he offers proof that Androtion dated the introduction of ostrakismos to 488/487, is by no means the only one possible (ibid. 394); and (b) a proper understanding of the purposes of the institution provides ... an explanation, by which the delay [between introduction and actual use] can be accounted for, in historical terms (ibid. 396).
As for Kagans first argument, one need only point out that after all everything is possible except that which is an outright and patent impossibility; that the traditional interpretation is possible; and that, by merely putting forward another possible interpretation, one does not prove the traditional interpretation impossible. Kagans historical arguments carry no weight in the present context. By setting out to prove that a date under which introduction and first application of the law on ostrakismos is historically impossible, he cannot assert with any justification that Androtion cannot have advanced such an impossible date: unless of course one were to adhere to some fundamentalist creed, ruling out on religious grounds that an ancient source may offer an account which flies in the face of historical probability. See p. 29; 34.
49 Most commonly one of three types: verbatim quotation, abridged quotation, adapted quotation (but not a true paraphrase). For an alternative hypothesis why AP and lexicographical sources produce nearly identical texts, see n. 16.
50 Firsts evoked the keen interest of ancient writers. Cf. A. Kleinguenther, *PRW=TOS EU(RETH/S: Untersuchungen zur Geschichte einer Fragestellung, Leipzig 1933 (Philologus, Supplementband 25, 1).
51 E.g., Hdt. 1.59.4-5, compared with AP 14.1. A line-by-line comparison, in the same way in which I compared AP with Harpokration, shows a significant qualitative difference: AP 14.1 is much farther removed from Hdt. 1.59.4-5 (whose name he actually cites at 14.4) than AP 22.4 is from the alleged Androtion of Harpokration. This example is closer to Herodotos than any other passage in AP which appears to be based on Herodotos.
52 Cf. Keaney, A narrative pattern in Aristotles Athenaion politeia Studies ... Sterling Dow . . ., Durham NC, 1984 (GRBS Monographs 10), 161-163, esp. at 163 fin.; also H. Ryffel, *M*E*T*A*B*O*L*H *P*O*L*I*T*E*I*W*N, Bern 1949 (Noctes Romanae 2), 23f.; 145; J. Touloumakos, Die theoretische Begründung der Demokratie, Athen 1985, 29ff.; 112; etc. (cf. my review AAHG 42  195ff.); J. P. Dolezal, Aristoteles und die Demokratie, Diss. Bonn 1973, Frankfurt 1974 (Studienreihe Humanitas), 69f. At pol. 1284b17ff., Aristotle comes very near to launching a summary defence of ostrakismos. J. Day-M. Chambers, Aristotles history of Athenian democracy, Berkeley etc. 1962 (University of California publications in history 73) (repr. Amsterdam 1967), 13ff., however, convinced as they are that Androtion dated the introduction of the law on ostrakismos to 488/487, take exactly the opposite view: Aristotle wishes to portray Kleisthenes as having given powerful encouragement to democracy, whereas Androtion wanted to deny that Cleisthenes had either founded radical democracy or introduced ostracism (ibid. 14).
Cf. Kagan (n. 46) 395: both Aristotle and Androtion knew that Kleisthenes had been the originator of ostracism and that Hipparchos had been the first man ostracized. They further agreed on the date of that ostracism and on the reason for the establishment of the law. As there is no disagreement between them, the ... argument [i.e., that Androtion disagreed with Aristotle on the date] against the Kleisthenic foundation of ostracism has no validity. Kagans conclusions were not accepted (as witnessed by the outpouring of articles on ostrakismos which are based on the premiss that Androtion dated the passage of the law on ostrakismos to 488/487). This may be partly because Kagan argues from possibilities (see n. 46), and because he thinks he can support his interpretation of Harpokration by an appeal to historical reasoning.
54 See p. 42 with n. 50. The length of the ostracisés term of exile is one such example (regardless of the historical facts; cf. p. 29). Diod. 11.55.3 and 87.1 gives it as five years from a tradition that surfaces in a fuller version in LexCantabr (cf. n. 11a) s.v. O)STRAKISMOU= TRO/POS = Philochoros FGrHist 328 F 30. Jacoby marks the initial two-thirds of the gloss as a verbatim quotation I would recommend a more cautious approach (cf. my similar cautioning regarding FGrHist 328 F 115, op. cit. [n. 5]).
Another example is the demarcation of the area the ostracisés had to observe after 481/480, AP 22.8; LexCantabr = FGrHist 328 F 30. Cf. Rhodes, Comm. AP 28.2, on the possible meaning of AP 22.8, and on various attempts at instilling meaning by means of emendation.
Irrespective of the historical facts behind all this, it is obvious that there was disagreement, and that some miraculous enlargement of the range of knowledge ensued in the process (cf.p. 36 and n. 34).