1. The most significant works applying narratology to Thucydides are Hornblower, S. (ed.), Greek Historiography (Oxford 1994)Google Scholar and A Commentary on Thucydides (2 vols., Oxford 1991–96)Google ScholarPubMed; Rood, T., Thucydides: Narrative and Explanation (Oxford 1998), esp. 11–14Google Scholar and 294–96; Gribble, D., ‘Narrator Interventions in Thucydides’, JHS 118 (1998), 41–67CrossRefGoogle Scholar. Like these writers, especially Rood and Gribble, I aim in general to defend the integrity of the text—to adopt a unitarian standpoint. Nevertheless to my mind some of the best older work on Thucydides has been done by ‘analysts’ such as Gomme, A.D., A Historical Commentary on Thucydides (Oxford 1945–81)Google Scholar, and Westlake, H.D., Individuals in Thucydides (Cambridge 1968)Google Scholar.

2. For an argument on stages of composition, see Gomme, A.D., ‘Four Passages in Thucydides’, JHS 71 (1951), 70–80CrossRefGoogle Scholar, though in his commentary he almost uniquely sees little tension between the two passages. For the possibility that this section was composed soon after the event, see Westlake, H.D., Essays on the Greek Historians and Greek History (New York 1969), 138–44Google Scholar. I am tempted by Westlake’s thesis that the ‘later war’ mentioned at 4.108.4 is the immediately subsequent campaign to recover the cities, and not the Ionian War. Thucydides tells us at 5.26 that in fact everything in the Peloponnesian War strictly speaking is part of the same war, and that one cannot really separate out its various stages, like those nowadays known as the ‘Archidamian War’ and the ‘Ionian War’. So at 4.108.4 he may be using the word more loosely to refer to a campaign, and there is no decisive reason to identify this campaign with the Ionian War.

3. Hornblower’s commentary (n.l above) covers much of the recent scholarship on this topic. He is more inclined to see stages of composition than I am; still, he posits that both passages are late and come from the same stage of Thucydides’ thought. In general, the unitarian scholars tend to argue that Thucydides has deliberately left a tension to make the reader think: e.g. Connor, W.R., Thucydides (Princeton NJ 1984), 130–35Google Scholar, and Rood (n.l above), 75, who sees a ‘suggestive tension between the two prolepses’ particularly in their representation of Athenian power and the consequences of Brasidas’ actions. On the other hand, analyst scholars see incompatible stages of composition: e.g. Westlake (n.l above) on Brasidas, and Andrewes in Gomme’s Commentary (n.1 above), v.364. The tension is not always expressed in these terms, but it seems to me that my formulation goes to the heart of the inconsistency between the two chapters.

4. Dewald, C.J., ‘Practical Knowledge and the Historian’s Role in Herodotus and Thucydides’, in Jameson, M.H. (ed.), The Greek Historians: Literature and History. Papers Presented to A. E. Raubitschek (Saratoga CA 1985), 47–63Google Scholar, groups together these masters of intelligent planning under the heading ; she argues that these men serve as models for historical analysis. This argument inspired me to write my dissertation: Watts-Tobin, A., ‘Generals and Particulars in Thucydides’ (Diss. Univ. of Southern California 2000)Google Scholar.

5. For the purposes of this study, I take ‘focalisation’ to mean more or less the same as ‘point of view’. I find the term useful because it establishes activity of focalising as a kind of surrogate narrative. When a character’s thoughts or words are presented in the text, that character takes over some of the functions of the narrator. The focalisation attributed to a specific character may well represent his/her values, interests and goals (e.g. Phoenix’ speech to Achilles in Book 9 of the Iliad). To recognise this point reminds us that the narrator too gives a personal perspective on events. At various points in the History, Thucydides delineates the perceptions of actors. For instance, in my translation of 4.81 above, I have highlighted the point that the results for Sparta were intended by Brasidas: Thucydides marks this out by using the infinitive-result construction . The more extended examples of focalisations, the ones which affect Thucydides’ judgments of responsibility and intelligence, involve accounts of the ideas those actors hold in mind—consciously or unconsciously—when they make decisions (e.g. 4.2.2 and 7.42.3). Such passages present some of the most interesting problems of focalisation in Thucydides.

6. Thucydides quite frequently refers to the general mood of the Greek world, and especially its attitudes about the impact of specific events on the Athenian empire. Some examples are 2.7–8, 2.65,4.55,4.65,4.108, 7.28,7.87, 8.1, and 8.27.

7. For important discussions of focalisation generally, see Genette, J., Narrative Discourse [tr. Lewin, J.E. (Oxford 1980 [orig. 1968]), 203–24Google Scholar; Bal, M., Narratology (Toronto 1985)Google Scholar; de Jong, I.J.F., Narrators and Focalizers: The Presentation of the Story in the Iliad (Amsterdam 1987)Google Scholar. Note that my proposed levels of focalisation are rather different from the ones proposed by Bal at 110–14. The main works that apply narratology to Thucydides are listed in n.1 above.

8. In what follows, I will be adopting the view that the speeches are substantially accurate versions of the speeches given by historical characters, and that they at least represent in some faithful detail the arguments the characters used. In fact little of the main argument depends on this assumption, but I do use it for illustrating different theoretical possibilities. For example, if we hold that the speeches have been chosen for inclusion because they contain arguments Thucy-dides thinks are important, then the speeches in the work may contain pieces of analysis that Thucydides has adopted for himself. Possible examples could be Perikles’ statement about Athenian originality at 2.41 and his arguments about civic responsibility at 2.60.4–7 (see n.30 below). Or it is possible that he has taken ideas, but not given a version of that speech in the work. For instance, though Thucydides states that Spartan fear of Athens as a reason for the war was ‘least mentioned’ (1.23, 88), it may be that one speaker did explicitly make this claim but most speeches (including the ones he gives) passed over it. Further, in this paper I am leaving aside the question whether we can separate the perspectives of Thucydides as narrator from those of Thucydides the author. I assume that the author can reasonably be identified with the narrator. For a rare objection to this see Syme, R., ‘Thucydides’, Proceedings of the British Academy abbr? 48 (1962), 39–56Google Scholar. To establish a significant distinction between these roles, one would need to demonstrate places where Thucydides gives or implies judgments that he does not believe himself.

9. This hypothesis forms a direct response to the still widely cited arguments by Hunter, V.J., Thucydides The Artful Reporter (Toronto 1973)Google Scholar, passim, and Schneider, C., Information und Ab-sicht bei Thukydides: Untersuchung zur Motivation des Handelns (Gottingen 1974), 127–37CrossRefGoogle Scholar.

10. For a fuller presentation of this argument, see the second chapter of Watts-Tobin (n.4 above). The conclusions in this paper form an argument by enumeration: that there is a body of evidence indicating that this hypothesis is correct.

11. See especially Stahl, H.-P., Thukydides: Die Stellung des Menschen im geschichtlichen Prozeβ (München 1966Google Scholar: the author’s English translation is to be published in 2003 by the Classical Press of Wales); id., ‘Speeches and Course of Events in Thucydides’, in Stadter, P.A., The Speeches in Thucydides: A Collection of Original Studies with a Bibliography (Chapel Hill 1973), 60–77Google Scholar; Edmunds, L., Chance and Intelligence in Thucydides (Cambridge MA 1975)CrossRefGoogle Scholar.

12. Gribble (n.1 above, 59f.) sets out the sequence and complexity of focalisation in this passage, stressing the way Thucydides alternates his own view and those of actors in the narrative. This alternation indexes the levels of meaning these events hold for the historian and for the participants, and contrasts story time and real time. I will focus on another use for it: the distinction between the reader’s retrospective view of the events with what was apparent at the time.

13. Gribble (n.1 above, 46, 53) discusses the placement of several of the narrator interventions in Thucydides, arguing that they are strategically arranged to draw the reader’s attention to what both Thucydides and the reader bring to the narrative. He argues that Thucydides draws attention to his role in both presenting facts and analysing them. Gribble shows how historical analysis involves not just judging actions but also presenting possible alternative consequences.

14. They send no support in the short term, and hardly any support ever materialises. The later narrative records that the Spartans in fact make two attempts to send reinforcements: 4.132.1 and 5.13.1. Both are turned back at Thessaly (which even Brasidas found difficult to cross—4.78.3). The fate of the first is mentioned in n.16 below, and the second founders because of Spartan disinclination to advance after the death of Brasidas. Gomme (n.1 above) ad 5.13.1 calls the Spartan response ‘pusillanimous’.

15. This judgment applies to the cities that revolted voluntarily. Straight after the victory at Amphipolis, Thucydides tells us, Brasidas makes a series of attacks on cities through the winter of 424/3. He starts with Akte, the Mt Athos peninsula, and attacks Sane, Kleone, Akrothoi, Olo-phyxos and Dion—Sane and Dion hold out. Then Brasidas turns his attention to the prize of the bunch: the city of Torone on Sithonia. None of these cities can be criticised for a decision to revolt that was forced upon them. However their fate does strengthen the initial perception of Brasidas’ invincibility. The group of ‘latecomers’ properly includes those cities that chose to revolt during this winter, while the Athenians and Spartans were negotiating the truce of 423 (4.117–18). The star members of this group are Skione and Mende.

16. Thucydides blames the lack of Spartan support on the Spartans at 4.108.7 (similarly in the cryptic 4.117.1), even though his later narrative suggests that Brasidas has contributed to it. On his way back from the campaign with Perdikkas at Lynkos (4.124–28), Brasidas allows his army to ravage Perdikkas’ country. It is this incident, Thucydides claims, that makes Perdikkas a mortal enemy of the Spartans; he then uses his contacts in Thessaly to stop the detachment of Spartan reinforcements from crossing Thessaly into Chalkidike, all except a small group of leaders (4.132.1). This hatred does not enter Thucydides’ judgments of responsibility at 4.108.4, since it starts after the revolts of Skione and Mende have taken place. However, considered next to 4.108.7 and 4.117.1 this account does seem to highlight Thucydides’ inclination to blame Spartan leaders rather than Brasidas.

17. The closest Brasidas comes is his claim that Athens is weak, and the false evidence of the incident at Nisaia (4.85.7).

18. Ron, M., ‘Free Indirect Discourse, Mimetic Language Games and the Subject of Fiction’, Poetics Today 2(2) (1981)CrossRefGoogle Scholar, 35f. This discussion is more familiar to classicists as quoted in Fowler, D.P., ‘Deviant Focalisation in Virgil’s Aeneia’, PCPS 36 (1990)Google Scholar, 45f. Ron does not discuss levels of focalisation, or even—in fact—focalisation at all. Instead, he discusses free indirect discourse, which is a category of narrative rather than focalisation. We could claim that Ron’s point here concerns levels of free indirect discourse. I avoid using that vocabulary for my discussion, however, since many of the passages I discuss incorporate direct speech. We would then have to posit the existence of ‘free direct discourse’ or speak of ‘levels of point of view’ to avoid using the terminology of focalisation.

19. Ron (n.18 above), 36. The phrases quoted above could in this instance also count as implicit embedded focalisation (2b in de Jong’s schema [n.7 above]). This accounts for why Fowler brings an example of free indirect discourse into a discussion of focalisation. Focalisation and free indirect discourse do not always correspond like this.

20. Note that the focalisation F3 does not correspond exactly with the original F2. It accepts F2 as true, but may in addition reinterpret it or direct it towards a particular goal: for the wife, to impress her husband; for the Chalkidians, to solidify their inclination to revolt.

21. Thucydides has described the strategic importance of Amphipolis from the Athenian point of view at 4.108.1; the rest of the passage assumes that the Chalkidians and the Spartans also recognise its strategic importance.

22. Thucydides apparently does not expect the historical actors to see what he claims at 4.108.7, that the Spartans are not willing to sponsor Brasidas’ expedition any further. This point only becomes clear later. Brasidas, for example, is still waiting for Spartan support in 4.123. Thucydides does not criticise Brasidas for his naiveté; he criticises the Spartans for failing to use a military opportunity. Brasidas would presumably have been aware of political opposition to him in Sparta (4.108.7). He seems to have assumed that since the Spartans have decided to send out this expedition, they would be willing to support it if successful. In any case, Thucydides makes clear that it is reasonable for Brasidas to think that Sparta will support him. Similarly, Thucydides judges that it is reasonable for the Chians to revolt in 411, because it is not clear at that stage that Athens has the strength to resist a mass revolt (8.24.5).

23. The relation between F1 and F2 does not imply temporal priority. It is eminently possible that characters have the idea first, and then Thucydides adopts it. That person or group may articulate it in a speech that reaches Thucydides’ ears or notebook, and then Thucydides adopts that idea as his own. I am tempted to think, for example, that some Athenians really did say words to the effect that the Spartans would have been just as hard on their allies as the Athenians have been (1.76.1), or that Perikles actually used the conceit that the empire is like a tyranny, and that Thucydides believes these speakers were right. Of course, while Thucydides illustrates these points in the narrative we cannot demonstrate definitively that these speakers said them first.

24. Examples of people swallowing Brasidas’ words whole (even against their better instincts) include 4.78.3, 4.85–88, 4.106,4.120.3–5.

25. Hornblower’s commentary (n.l above), 86–88; Watts-Tobin (n.4 above), 225–27.

26. See Watts-Tobin (n.4 above), ch.3.

27. Many scholars have drawn attention to the way Brasidas misrepresents the situation at Nisaia (4.85.7, cf. 4.108.5). At Akanthos, Brasidas suggests that the Athenians are frightened of him because they did not want to attack when they had the chance to do so at Nisaia. But at 4.108.5 Thucydides recalls the crucial fact that Brasidas had a larger army in that engagement. In my opinion Thucydides is not suggesting that everything Brasidas says is false, since his audience must believe that his words are true. Compare Hunter, V.J., Past and Process in Thucydides and Herodotus (Princeton 1982), 147CrossRefGoogle Scholar. Instead, he gives an artful mixture of truth and falsehood.

28. As Perikles puts it (1.140.1): ‘For sometimes the course of things is as arbitrary as the plans of man; indeed this is why we usually blame chance for whatever does not happen as we expected.’ He implies here that the plans of man make outcomes arbitrary. They run they ‘haven’t read the book’ about how they are supposed to run (oral witticism by C.B.R. Pelling at a University of Pennsylvania colloquium, Nov. 2002).

29. In a contest where one side has decided not to provide opposition, the enemy has a free ride. Action in war is predictable if not contested (by humans or by fortune). The character has known goals and intentions, and will naturally pursue them. Once any obstacles to their achievement are removed, then the character would be expected to pursue them. In this sense, their action becomes predictable.

30. In his final speech, Perikles makes the point that it is the people’s responsibility to decide on the right advice from the several proposals delivered by speakers in the assembly. The adviser should not be held accountable for a bad decision even if he recommended it. In a dazzlingly bold argument (2.60.4–7), he asserts that the Athenian people should run to help the city and not ‘…blame me for having counselled war and yourselves for having voted it. And yet if you are angry with me, it is with one who, as I believe, is second to no man either in knowledge of the proper policy, or in the ability to expound it, and who is moreover not only a patriot but an honest one….So that if you thought that I was even moderately distinguished for these qualities when you took my advice and went to war, there is certainly no reason now why I should be charged with having done wrong.’

31. 2.84.2–3. Compare the less expert Spartans’ interpretation of the same event: ‘there were some chance events that worked against us’ ( 2.87.2); see also the chance wind that carried Demosthenes to Pylos and the accidentally sparked fire on Sphakteria (4.3.1, 30.2). Thucydides appears to consider the wind a matter of chance if it could not have been reasonably predicted by even those alert to weather patterns. In the same way, a character’s intentions and goals ought to be a known quantity; if there is no opposition to them then their pursuit of those goals can be called predictable. Action that does not fit those known principles and intentions may be considered arbitrary and unpredictable.

32. Hart, H.L.A. and Honoré, A.Causation in the Law ² (Oxford 1985), 136–42CrossRefGoogle Scholar.

33. Three ways in which legal thinking twists analyses of causation in its own fashion may be mentioned right away. First, legal theory concentrates on applying general principles to particular circumstances. It rarely concerns general issues per se. Second, legal cases tend to concern cases (and therefore causes) of harm rather than any other type of consequence. Third, and perhaps most crucially, it is very hard to use decisions in single cases, or small groups of cases, to arrive at a principle of responsibility; a host of other factors affect the actual sentence, such as the appropriateness of the punishment, the attempt to make the punishment represent the defendant’s responsibility for the crime, and the potential deterrent effect in setting a precedent. On the other hand, fortunately for us, but unfortunately for the people involved, legal cases abound: we have cases and decisions which represent exactly the possibilities I have mentioned above. There are many cases of a man leaving a pit in a road, and another man either falling into it or being pushed into it.

34. Hart and Honors have a very clear agenda in their work, which is to establish that legal conceptions correspond with ordinary people’s thinking on the matter: the views of what Lord Bowen called the ‘man on the Clapham omnibus’. This idea is quite unpopular nowadays. ‘The ordinary man’ is very much an artificial construct; many legal theorists however maintain that it is necessary and inevitable in discussions of causation. J.L. Austin is one of the wittiest and most scathing critics of this conception in philosophy, memorably describing it as a philosophical ‘own goal’. See Austin, J.L. and Warnock, G.J., Sense and Sensibilia (New York 1962), 6Google Scholar and passim, against especially Ayer, A.J., The Foundations of Empirical Knowledge (London 1958)Google Scholar.

35. Feinberg, J., Doing & Deserving: Essays on the Theory of Responsibility (Princeton 1974), 153Google Scholar, doubts the application of this principle. He gives some counter-examples to illustrate his objections. The following one he styles ‘The foolhardy bank teller’, and is roughly based on the case Noll vs. Marion, 347 Pa. 213 (1943). Imagine that Jones is waiting in line at a bank when a robbery takes place. The robber warns, ‘If anyone moves, I’ll shoot!’ The bank teller grabs something and dives to the floor. Then the robber shoots at the teller, misses, and the bullet ricochets and hits Jones, causing him injury. He then sues the bank because its employee created an unreasonable risk of harm by his action. In this case, Feinberg holds that the robber’s shooting counts as a ‘voluntary intervention’ intervening between the teller’s risky behaviour and the eventual harm. He argues that the teller’s action could reasonably be considered the ‘cause’ of the harm. This means that causation is not negatived by voluntary intervention in this case. This argument, like all his objections to Hart’s and Honor’s thesis, depends first on what counts as a voluntary action: he assigns voluntary intervention status to the shooting itself, and not to the teller’s foolhardy action. Second, the example depends on how consequence is distinguished from action. See Austin, J.L., How to Do Things with Words (Cambridge MA 1962), 107–20Google Scholar, for a discussion of how this latter problem applies to speech-acts (especially the distinction between illocutionary and perlocutionary acts). I do not find that Feinberg’s objections impinge on the distinctions Thucydides makes: his main distinction lies between action that could be considered arbitrary and unpredictable and action that is dictated by the circumstances. In Feinberg’s example, I think Thucydides would place the robber’s shooting in the latter category: he has already threatened to do just that. Action that is normal in the circumstances would not count as voluntary action. To return to Hart’s and Honor’s conception: when using the voluntary intervention principle, responsibility is traced as far back as the last voluntary act. This chain is intact so long as everything, including human actors, is functioning normally. The chain of causality is broken only by an arbitrary and voluntary human action.

36. We may compare the words of Wesley Snipes to Woody Harrelson in the movie White Men Can’t Jump. In the movie, Harrelson is addicted to pickup basketball, and his obsession threatens to tear apart his relationship with Rosie Perez. Meanwhile she has her own ambitions to be a professional game-show contestant, and threatens to leave him if he continues. The predictable thing happens. Snipes cajoles and persuades Harrelson to play one last, potentially lucrative, game. Harrelson plays and Perez leaves him. Harrelson goes back to Snipes and complains: ‘But you talked me into it!’ Snipes replies with one simple piece of advice: ‘Always listen to the woman.’ Snipes argues that he just made a proposal, and Harrelson accepted. His acceptance counts as a voluntary action.

37. Hermokrates uses this principle at 4.61.5: ‘There’s every excuse for the Athenians to be aggressive; I don’t blame those who wish to rule, but those who are over-ready to subject themselves.’ (Cf. also 6.76.4, 77.1.) This distinction between candidates for blame does not depend on the main point I am exploring in this paper; however, it does illuminate one important underlying theme: how assignment of blame depends on distinguishing ‘normal’ from ‘exceptional’ action.

38. Here Thucydides uses the umbrella term ‘the allies’ without acknowledging that some allies avoided making this decision, and for this very reason. The Mytileneans at 3.11 describe how they have been threatened by gradually expanding Athenian power even though they have retained possession of their fleet.

39. Dionysios of Halikarnassos (de Thuc. 18.7) and Marchant, , Thucydides Book VI (London 1897)Google Scholar, ad loc, comment on the artificiality of this sentence. Denniston, , Greek Prose Style (Oxford 1950), 13Google Scholar, observes that Thucydides generally eschews the sophistic parallelisms typical of Korax, Tisias and Gorgias, but notes certain exceptions. He describes a similar passage (4.61.7)—also spoken by Hermokrates—as ‘an unabashed bending of the knee to Baal’. The current passage would seem to be another exception: I must say I rather like it.

40. One of the more interesting examples is that of Phrynichos in 8.50, who saves himself by giving the enemy vital information.

41. We may compare this mastery with an anecdote about Pablo Picasso. He would always pay for everything by cheque. Picasso realised that his signature would be far more valuable than his payment for whatever item he purchased; consequently, retailers never cashed his cheques, and he bought things for free.

42. I have argued that Thucydides expects actors to look at the future bearing two particular considerations in mind: first, that they work out a view for themselves and not simply adopt someone else’s; and second, that that they look at how action is likely to affect other people’s interests and their own, given what is predictable about the world. The chain of predictability and responsibility follows those interests—unless xtixri intervenes either in human or non-human form. For more on this see the third chapter of Watts-Tobin (n.4 above).

43. Special thanks for help with this paper go out to the anonymous referee for this journal, who made many invaluable suggestions and pointed out one rather spectacular error; to Carolyn Dewald, who directed the thesis that underlies some of this material; to Dan Tompkins, who made time for me to supervise many stages of this paper; to Paul Crowe, who discussed the legal parts of the paper with me; and finally to Jamie Roscoe, Lois Risling and Zo Devine of the Center for Indian Cultural Development at CSU Humboldt, an idyllic place where most of this paper was written. Also, I have gained much from my conversations with Edwin Carawan, Carolyn Higbie, Martha Davis, Stephanie Schull, Tom Seifrid, and Charles Miller of the USC Law library and the Ham Fighters. Final thanks go to my wife, Jeanette Cooper, for love, support and a keen eye; and to the Pacific Rim Roman Literature Conference (especially Tony Boyle and Martha Malamud) for bringing us together.