I was stunned by my own ignorance when I did some quick research into the ancient sources for a crucial aspect of the Athenian trittyes reform, for a review of Siewerts book2 on the topic. AP 21.4, I discovered, is completely without parallel in the sources, both literary and epigraphical:
He (i.e., Kleisthenes) distributed the country (i.e., Attika) throughout the demoi amongst thirty parts, ten from the city and surroundings, ten from the shore, and ten from the interior, and those he called trittyes and he allotted by sortition three to each phyle, in order that each (i.e., phyle) share in all places. (AP 21.4)3
Thus it was not until after the speedy publication of PLond (BM 131) in 18914 that modern trittylogy could seriously get under way.5 The learned Busolt6 could still write in 1885: Aus je zehn Demen bildete Kleisthenes die zehn neuen territorialen Phylen (my italics). ...aber auch... gelegentlich ein Bezirk Enklaven (my italics) in einem anderen hatte, und die politische Karte Attikas mindestens ebenso bunt aussah, wie die Thüringens in Deutschland; and, in the preceding paragraph: ...die Einführung der dreissig Trittyen, die in perikleischer Zeit und späterhin lokale Aushebungsbezirke für die Flottenmannschaften waren und danach die Matrosendivisionen sowie auch wohl die Bataillone der Phylenregimenter bildeten, würde erst später erfolgt sein.
I was stunned by my ignorance because, until my discovery, I had been teaching for many years as a straightforward fact the model so neatly illustrated in every textbook and scholarly publication up to Traill and Whitehead.7 Never has this universally accepted truth
been challenged not even by the astute Beloch, although he came close to it when he wrote about the allotment of trittyes to phylai that ...ich ... die ganze Losgeschichte nicht glaube, and grandiosely reassigned the entire phylai reform to Peisistratos.8
The traditional textbook map of Attika in the much perfected version of Traill (see fig. 1, p. 27 below) used AP 21.4 as a foundation; the outlines of the three regions are then defined with the aid of the bouleutic inscriptions9 (which allow us to assign the demoi to their appropriate phylai), and by taking deme contiguity into account. A cluster of neighbouring demoi is regarded a trittys which can be easily labelled with the AP terms. A demos which does not neatly conform with this scheme and thus with AP finds itself attached, as an enclave, to some distant trittys of the same phyle.
This procedure hardly qualifies as scientific method and can only be understood as the result of a historical growth process itself. a consequence not least of the descent of Greek History from Greek Philology. The results produced by this method are quite incongruous.
(1) That stretch of coast which lies between the south and the west of the city, about one-sixth of the total shore line of Attika, is not at all *P (PARALI/A, PA/RALOS GE= , etc.; paralia; coast) but *A (A)/STU; asty; city).
(2) *P falls, as a consequence of (1), into two sectors, if not three. These are the Thriasion pedion10 to the west; the coast from the foot of Mt. Hymettos to the Laureion, the entire Laureion itself, the east coast to as far north as the foot of Mt. Brilessos/Pentelikon; and, probably separated from it,11 the Marathonian plain and Rhamnous.
(3) Except for Thorai, Anaphlystos, Atene, Thorikos, Potamos Deiradiotes and Deiradiotai, the other shore demoi in the Laureion are not at all coastal in real terms; nor are Oinoe (Hippothontis) and Phyle, nor Upper Lamptrai. At least 40 councillors.
(4) Enclaves. (On this, see pp. 29ff. below).
(5) The trittyes thus created range in their bouleutic representation from 9 to 27 councillors (18% to 54% of the total contingent of 50 councillors per phyle).12 Even if trittys
excessively unbalanced. The trittyes served purposes other than merely decorative ones. Just as the old name phyle was used again to serve as the term for one tenth (exactly ten times 50 bouleutai), so too the old name trittys was used again to serve as the term for the phyles subdivisions of which there were three. The people assembled by phylai for at least some of the business of the assembly (e.g., ostrakismos), and when they went into battle. The military aspect in particular requires phylai of fairly equal size. By extension, its military subdivision too had to be of fairly equal size.14 Something closer to an arithmetical third surely was required. If a military trittys was approximately a third in arithmetical terms, then surely the bouleutic trittys too will have represented something very nearly one third in arithmetical terms. Since fifty bodies cannot (at least not until the advances of the Christian Middle Ages in the quartering of bodies) be neatly thirded, only an approximation can be achieved, which ought to be 17+17+16.
Thompson15 found exactly these groups of twice 17 and once 16 in the bouleutic lists. The lists from the period 408/7 to 321 provide enough evidence to declare this a fact. The fact of these thirds is there for everyone to verify, and there is even indisputable evidence that they were called trittyes.16 There is no evidence before 408/7 from inscriptions but Traill17 has speculated that these thirds may in fact reflect the original organization, both political and military, of Kleisthenes. He need not have been that cautious: he is stating a fact. There is no indication anywhere in our sources of a reform which would have introduced these thirds in addition to different (unequal) thirds. Reform that was capable of creating ten fairly even sized phylai surely was capable of creating similarly even sized subdivisions of these phylai at the outset.
There is admittedly one potential spoiler. The demos of Akharnai is assigned no fewer than twenty-two bouleutai, i.e., five or six more than a third. There is no indication in the inscriptions that, or how, these twenty-two might have been subdivided. Eliot18 thinks that this suffices to refute the notion of arithmetical thirds. I would rather view the case of Akharnai as the exception that proves the rule. Akharnai was, after all, an immensely populous and powerful demos.19 Most important, it is one exception, in contrast to the zoo of
exceptions which we are about to encounter in the next paragraphs under the AP scheme.
Now, Siewert, it seems to me, must, at an early stage of his study of the organisation of Attika, have drawn a map which combined both Traills map 2 and the logical consequences of the evidence of the bouleutic inscriptions. If one does this (as I have done; I regret that I cannot reproduce my map here because it requires three-colour print), one obtains a most startling result: it cries out for an explanation.20 For what previously looked like a neatly ordered arrangement suddenly turns into a strange mosaic of enclaves. So strange in fact that Siewert found himself compelled to distinguish between three categories of enclaves which he labelled regionale Enklaven, außerregionale Enklaven, and Quasi-Enklaven". The außerregionale enclave is perhaps the most straightforward case, together with its curiously named cousin Quasi-Enklave." These are demoi which are located in a region which is not the same, in AP terminology, as that of the other demoi of the same trittys. Probalinthos, for instance (and if it has been correctly placed at the eastern foot of Mt. Brilessos), is located in the *P but surrounded by demoi of other phylai, and it is attached to the A demos and trittys of Kydathenaion, thus representing an außerregionale enclave. Rhamnous, on the other hand, although not physically separated from other demoi of the same phyle, is not attached to its neighbouring *P trittys but to the *A demos Phaleron (we do not know the trittys name). The enclaves within theAP system are illustrated in fig. 2 (p. 30 below).
Außerregionale or Quasi-Enklaven afflict every region. *A has two that go with *P and one that belongs to *M (one bouleutes each). *M contains seven demoi that are *A (fourteen councillors). *P includes three *A and three *M demoi (with fifteen and sixteen bouleutai, respectively). Thus 48 (of a total of 500) from sixteen demoi are proven enclavists of this sort. In the category of regionale enclaves, which means that a demos is in the right region but still in the wrong place because the neighbours are from other trittyes or phylai, there are (on the basis of Siewerts map 4) 11 demoi of *A or *P trittyes (31 councillors); of these, ten (with thirty bouleutai) are in Leontis (perhaps offering one of the clues to the origins of the new phylai system, cf. p. 33 below).
There are then 79 bouleutai (about one sixth of the total) from twenty-seven demoi (c. one fifth of the total number of demoi), who are confirmed enclavists in one or another of the three Siewertian categories for enclaves. Is it possible to accommodate this very large deviation from what might be termed the norm? Can the map of Attika be redrawn to accommodate the principle suggested by AP 21.4 and the evidence of the inscriptions? My fig. 3 (p. 31 below) is such an attempt. This map is so obviously bizarre that it will be dismissed. It is not, however, any more at variance with reality or, as some might say, common sense and logic, than Siewerts map 4, Traills maps 1 and 2, and all other maps which are based on AP 21.4. They must all be regarded as invalid as my fig. 3 (p. 31 below).
In an attempt at saving tradition, Siewert invented the Zentralwegprinzip. Demes are strung out along Zentralwege, trittyes are formed by stringing them together along these routes, which one would have followed into Athens in case of a military crisis. This solution requires us to assume that only military considerations, most applicable to offensive warfare, determined the forming of trittyes; it is predicated on a knowledge of late archaic and early classical traffic arteries in Attika which we do not possess.21 Siewert, however, still largely remains on the solid ground of the literary, epigraphical and physical evidence which are the foundation of scholarship. The same cannot be said of those who use the chaotic evidence
described in the preceding paragraphs to read the mind of Kleisthenes.
I would suggest that only the application of basic scholarly method will further the cause of rational historiography. By reviewing the sources we have seen that there are two classes of evidence; they do not produce identical information. On the one hand there is the literary evidence of AP, on the other we have the epigraphical evidence of the official documents in Agora 15. It is the classic dilemma that the historian encounters all too frequently. The solution can only be one of three: reconciling the contradictory evidence; dismissing the inscriptions in favour of AP; or dismissing AP in favour of the inscriptions.
The writing proceeds at three levels in AP 21.4: (1) easily observed facts are stated; (2) names are attached to these facts; (3) a theoretical explanation for the facts is provided.
(1) The facts are, even to the superficial observer who has no special knowledge of the
intricacies of Athenian government, that in every phyle there are men from often distant places in Attika, and members of different trittyes tend to come from different parts of Attika.
(2) The names given to the different and distant Attic places, in order to satisfy a desire for some sort of order, are common terms which can be used anywhere in Greece to describe the make-up of any typical Greek state that has access to the sea. Korinth or Sikyon could be described by using the words A)/STU, PARALI/A and MESO/GEIOS, or it could be Miletos or Syrakousai, etc.22
(3) The theory by means of which the facts can be suitably interpreted derives from Aristot. Pol. 1319b. After having reviewed the characteristics of demokratia he turns to its four types, in descending order. The lowest category includes the requirement of new phylai whose total number is to be greater, and all must be mixed with one another, for which democratic Kyrene is cited as an example, as well as the demokratia of Kleisthenes. This theory, conceived and expressed many years before AP, serves in AP to provide the theoretical foundation.23
Of these three levels, only (1) qualifies as evidence. At this basic level, however, the authors knowledge is hardly significant: it only becomes clear that he knew of the presence of men from a great many different places in the phylai and of trittyes forming these phylai which too came from different parts of Attika. No precise information can be gleaned from this that would either add to or contradict the evidence of the inscriptions.
The inscriptions, on the other hand, offer primary evidence in the truest sense of the word. They present official documents, produced by Athenians for Athenians. The names of hundreds of Athenian citizens who had duly passed dokimasia and euthyna vouch for them. These documents prove that each phyle consists of three trittyes;24 that each trittys, like the phylai, has a name; that each trittys contains 16 or 17 bouleutai; that a trittys may or may not
consist of demoii which are immediate neighbours; that there is not one single phyle which forms a territorial block; and that all phylai are made up of demoi located in different parts of the country.25 The inscriptions do not tell us how this peculiar system was generated, nor why. I have done only preliminary work on those questions. I would, however, venture to predict that the how will reveal not a glorious manifestation of a genius at work but, rather, a process best described as muddling through. The why will find its explanation (emphatically denied by AP 21.3) in the previous existence of twelve trittyes (making up the old four phylai). In contrast to the council of 400 or the board of ten generals which, at the time they were created, were without precedent or antecedent and could therefore be shaped without reference to existing bodies, such was not the case when it comes to the Kleisthenic ten phylai and thirty trittyes. Here the old system must be taken into account and we must attempt to determine if and how the old and the new systems relate to each other.
Had the reform of the phylai started from a tabula rasa, something along the lines suggested by AP 21.4 might have been conceived. The inscriptions prove it was not unless of course someone were to suggest that subsequent reform turned an originally neat system into that almost crazy one which, as Siewert aptly observes, could only with the aid of the computer be tested for randomness or planning.26 Or unless one delights in the writing of fiction posing as scholarship, uncovering the most secret devious schemes of the political genius Kleisthenes. If history is a science, however, the historian can do no more than, first, assess the genesis and value as evidence of each source, and, second, reconstruct what can be reconstructed on the basis of the best evidence available or, if no such evidence exists, say so. In our case, fortunately, evidence of the best order exists in the form of state documents on stone. AP 21.4, however, which had succeeded, merely by virtue of its antiquity and its association with the name of Aristotle, in invading, usurping and obliterating the critical spirit if not historical competence of generations of scholars and a century of scholarship, ought to be buried once and for all.27
1 This little paper is meant to stimulate discussion, which, I believe, the editors of this meritorious new publication intend (although I cannot predict my success). I have therefore kept annotation to a minimum, and because P.J. Rhodes, A commentary on the Aristotelian Athenaion Politeia (Oxford 1982) can be relied on to provide the reader with most of what he needs here. The main thrust of this paper is identical with that of short communication read at the CAC meetings, 1986 06 01.
4 Rhodes, CAAP, 3-5.
5 Siewert appropriately begins his brief survey (3ff.) of the history of Trittyenforschung with the year, 1891 and the publication of AP. Knowledge about the trittyes was previously limited to largely antiquarian information from lexika and skholia, often contaminated to the point of corruption. It was known that AP had something to say about the trittyes of earlier ages (preceding the present ch. 1), F 385 Rose and the testimonia listed in the Teubneriana of AP at 9.3.
6 G. Busolt, Griechische Geschichte bis zur Schlacht bei Chaironeia, 1. Theil (Gotha 1885) 614. At 614, n. 5, he observes that eine solche Karte fehlt noch (see text below). See pp. 29ff. below for the question of the Enklaven.
7 J.S. Traill, The political organization of Attica: a study of the demes, trittyes, and phylai, and their representation in the Athenian council (Princeton 1975); D. Whitehead, The demes of Attica, 508/7-ca. 250 B.C.: a political and social study (Princeton 1986). Traills map 2 (dark grey for *M, light grey for *P, white for *A; red, blue, and yellow in the separately sold version) is reproduced without alteration by Whitehead, p. xxiii, even though he advertises (xxii) that it stands in need of some revision (which Traill himself plans to undertake), but it remains the best that we have at present; see fig. 1 below.
8 K.J. Beloch, Griechische Geschichte, 12, 328ff. (quotation from p. 330).
9 Until we have IG II3 for these inscriptions, see The Athenian Agora 15 (henceforth Agora 15) Inscriptions; the Athenian councillors, ed. B.D. Meritt and J.S. Traill (Princeton 1974). The earliest list is one from Erekhtheïs from 408/7 (Agora 15, 1), all others of the pre-Macedonian era are later.
10 For these terms (and divisions) cf., e.g., Thouk. 2.19.2, 55.1.
11 The location of Probalinthos between Marathon (Aiantis) and the demoi of Aigeïs to the south is not absolutely certain, according to Traill and Siewert (maps).
12 In Traills reckoning, the trittyes of each phyle add up as follows: Erekhtheïs, 16+22+12; Aigeïs, 11+14+25; Pandionis, 12+19+19; Leontis, 13+19+17 (?); Akamantis, 14+17+19; Oineïs, 11+17+22; Kekropis, 15+15+20; Hippothontis, 20+20+11 (?); Aiantis, 9+25+16; Antiochis, 10+27+13. (In two instances, Traills partly hypothetical figures do not add up to fifty. The trittyes are given in the order *A+*P+*M.) For what it is worth I add the totals, for each region: 131,195,174 (26.2% + 39% + 34.8%). *A trittyes range in size from 9 to 20 with an average size of 13.1; *P trittyes, 14 to 27 (19.5); *M trittyes, 11 to 25 (17.4).
13 C.W.J. Eliot, Aristotle Ath. pol. 44.1 and the meaning of trittys, Phoenix 21 (1967) 79-84.
14 Cf. Siewert, 141ff.
15 W.E. Thompson, *TRITTU=S TW=N PRUTA/NEWN, Historia 15 (1966) 1-10. Whether these are the TRITTU/ES TW=N PRUTA/NEWN of AP 44.1 as Thompson thought I do not know; I do not wish to touch this subject here. It does not in an essential way affect our argument; the term itself may well be a coinage of AP rather than one current in Athenian everyday political jargon.
16 Agora 15, 26,14; cf. Siewert 14f.; J.S. Traill, Diakris, the inland trittys of Leontis, Hesperia 47 (1978) 89-109, at 90, l. 52 (a 371/0); Rhodes, CAAP at 44.1f. for further literature and discussion of the TRITTU/ES TW=N PRUTA/NEWN of AP 44.1.
17 J.S. Traill, Hesperia 47 (1978) 109.
18 C.W.J. Eliot, Phoenix 21 (1967) 85f., n. 18. For Oineïs we have only one well preserved list and two badly damaged ones (Agora 15, 17; 43 and 48).
19 Thoukydides (2.20.4) could get carried away into writing that they were able to put 3,000 hoplites into the field.
20 Although about two dozen demoi have not yet been fixed on the firmament of the map of Attika, we may safely disregard these planetai because they are mostly small (at best three councillors) and their status would not significantly change the picture (if anything, it might make it even more chequered).
21 Traill (in private conversation some time ago) appears sceptical about Siewerts Zentralwege.
22 The terms PARALI/A, PA/RALOS GH= and the like occur in both Hdt. and Thouk. The stretch of coast from Phaleron to the Laureion as the scene of certain action is meant by Hdt. 5.81.3; Thouk. 2.55.1, 56.1, 56.3. Upper and Lower in place names, normally expressed by U(PE/NERQEN and KAQU/PERQEN, is given as *LAMPTRH=S PA/RALOI / KAQU/PERQEN in Agora 15, 14, 34-35 (first published by W.K. Pritchett, Hesperia 11  231-239). KAQU/PERQEN is treated as a synonym of MESO/GEIOS by Hdt. 4.100.2 (on Skythia). E)S TH\N MESOGAI/AN is the direction in which the Persians pursue the Athenians of the faltering centre at Marathon, Hdt. 6.113.1. There are many instances of MESO/GEIOS meaning simply inland in both Hdt. and Thouk., in Greece, Ionia, Libya, Egypt, etc. (There is, incidentally, a cult association of the *MESO/GEIOI with a Herakleion at Kholargeis, IG II2 1244-1248.) A)/STU occurs with expected frequency, meaning what one would expect it to mean, in both authors, often alternating with its synonym PO/LIS. Both terms always refer to the city proper, and never to the surrounding PEDI/ON as well, in Thoukydides accounts of the invasions of Attika (e.g., 2.19ff., 55f.; 7.19.1f.).
24 Significantly, there is no hierarchy of trittyes within each phyle. For instance, four different ways in which the trittyes of Pandionis are listed are attested: Paiania, Myrrhinous, Kydathenaion (Agora 15, 10; 32., 47); P., K., M. (15, 26; 42); M., K., P. (15, 12); K., P., M. (15, 15).
25 There is a wide range, from the nearly territorial Aiantis (Aphidna, 16; Marathon, 10, Oinoe, 4, Trikorynthos 3; Rhamnous, 8 41 councillors or 82%; only Phaleron, 9, spoils this picture), to Leontis, in which hardly any demoi are neighbours (30 bouleutai from Siewerts regionale Enklaven; see p. 29 above). No phyle, incidentally truly fits the AP scheme *A+*P+*M, although Pandionis perhaps comes closest.
26 Siewert, 122, n. 174.
27 Addendum: a prospectus announcing J.S. Traill, Demos and trittys: epigraphical and topographical studies in the organization of Attica (Toronto: Athenians, Victoria College, 1986), pp. viii + 151, 16 plates, 5 maps, reaches my desk as I complete the proofs of this paper. KHK [cf. my reviews in Gymnasium 96 (1989) 170f.; CR ns 39 (1989) 67-69; and my article On the consequences of following AP 21,3 (on the phylai of Attika) in Chiron 19 (1989) 347-365.]