1. I am following the syntactical interpretation of Erbse, H., ‘Der erste Satz im Werke Herodots’, Festchrift B. Snell (Munich 1956), 209–22Google Scholar, esp. 217–19. See also Krischer, T., ‘Herodots Prooimion’, Hermes 93 (1965), 159–67Google Scholar; Hommel, H., ‘Herodots Einleitungsatz: ein Schlüssel zur Analyse des Gesamtwerks?’, Gnomosyne: Festchrift W. Marg (Munich 1981), 271–87Google Scholar.

2. There is no basis for restricting a priori the meaning of erga megala te kai thōmasta, and the text of the Histories does not support such view. For a history of the discussion of this phrase in the proem, see Barth, H., ‘Zur Bewertung und Auswahl des Stoffes durch Herodot (Die Begriffe thōma, thōmazō, thōmasios und thōmastos) ’, Klio 50 (1968), 93–110CrossRefGoogle Scholar, who also cites all the ethnographical facts qualified with thōma and derivatives and some of the narrative facts as well. A list of the latter group can be derived from the occurrences of thōma etc., in Powell, J. E., A Lexicon to Herodotus (Cambridge, UK 1938 Google Scholar), s.vv.

3. For anticipatory alios in Herodotus as marker of temporary selection, see Munson, R. V., Transitions in Herodotus (Diss. Pennsylvania 1983), 81–86Google Scholar; Wood, Henry, Herodotus: An Analysis of the Formal Structure (The Hague 1983), 14 Google Scholar.

4. By ‘insertion’ I mean a passage formally subordinated to the narrative which it interrupts. I avoid the term ‘digression’ because it has been applied differently by different critics, and sometimes used to designate a passage subjectively regarded as secondary in substance. See especially Jacoby, Felix, art. ‘Herodotus’, RE Supplementband 2 (Stuttgart 1913), 205–520Google Scholar, esp. 284, 287–90, 380–85; Pagel, K. A., Die Bedeutung des aitiologischen Moments für Herodots Geschichtsschreibung (Diss. Berlin 1927), 41–46Google Scholar, 54–55, 58–59; Erbse, Hartmut, ‘Tradition und Form im Werke Herodots’, Gymnasium 68 (1961), 239–57Google Scholar, esp. 243–57; Cobet, Justus, Herodots Exkurse und die Frage der Einheit seines Werkes (Wiesbaden 1971 Google Scholar), passim, and esp. 68, for the author’s definition of ‘digression’.

5. Pagel (n.4 above), 41–61, rightly underlines the impulse to explain in the Histories, but as he distinguishes ‘episodische’ from ‘motivierende Exkurse’, he tends to do so on the basis of contents, disregarding the explanatory justification expressed in the transition. This leads him to identify even the flashback on Cyrus (1.95–130) and the Gyges–Candaules episode (1.13–17) as episodic digressions.

6. ‘Flashforwards’ are less numerous and tend to be considerably shorter than ‘flashbacks’. See e.g., 6.72; 7.137, 151. Narrative insertions whose relative time with respect to the surrounding narrative remains undetermined are very infrequent. See e.g. (besides ‘Arion’, pp. 9f. below) 4.143–44.

7. Immerwahr, Henry, Form and Thought in Herodotus (Cleveland 1966), 145–47Google Scholar and n.191; Krischer, T., ‘Herodots Schlüsskapitel, seine Topik und seine Quellen,’ Eranos 72 (1974), 93–100Google Scholar. On the thematic relevance of this story, see also Ayo, N., ‘Prolog and Epilog: Mythical History in Herodotus’, Ramus 13 (1984), 31–47CrossRefGoogle Scholar, esp. 39–42.

8. Pagel (n.4 above), 43, identifies it as the ‘Musterbeispiel’ of episodic digression. This insight with regard to the uniqueness of this passage makes his assigning it to the same category as the Gyges episode (see n.5 above) all the more unconvincing.

9. Flory, S., ‘Arion’s Leap: Brave Gestures in Herodotus’, AJP 99 (1978), 411–42Google Scholar, esp. 412. The opposite view, maintained by Erbse (n.4 above), 250, and by Myres, J. L., Herodotus: Father of History (Cambridge, UK 1962), 83 Google Scholar, has no support in the text, since Herodotus carefully waives the opportunity to mention actions plausibly taken by Periander (e.g., the punishment of the sailors).

10. See Jacoby (n.4 above), 381 and 388f. When this tendency is not followed, a looser structure results. This happens with some of the great ethnographies (e.g. 1.92–94, 131–40) and, among narrative insertions, only in the case of the Artembares anecdote (cf. p.7 and n.7 above). The Cyrus flashback is a notable case of postponed insertion (see 1.75.1), but the explanatory introduction at 1.95 firmly anchors it to the ‘main narrative’.

11. Shimron, B., ‘ Prbtōs tōn hēmeis idmen ’, Eranos 71 (1973), 45–51Google Scholar.

12. Flory (n.9 above), 419. The origin of the tradition is discussed by Bowra, C. M., ‘Arion and the Dolphin’, MH 20 (1963), 121–34Google Scholar.

13. Cf. 3.9.2; Pearson, L., ‘Credulity and Skepticism in Herodotus’, TAPA 72 (1941), 335–55Google Scholar, esp. 335; Groten, F. J., ‘Herodotus’ Use of Variant Versions’, Phoenix 17 (1963), 79–87CrossRefGoogle Scholar. Herodotus’ ‘criteria of truth’ have been recently discussed by Hunter, Virginia, Past and Process in Herodotus and Thucydides (Princeton 1982), 50–92CrossRefGoogle Scholar; see also Waters, K. H., Herodotus the Historian: His Problems, Methods and Originality (London and Sydney 1985), 34–39Google Scholar.

14. E.g. 1.70.2–3; 4.173; 5.44–45; 6.137; 8.84, 94.

15. E.g. 3.2–3, 56; 5.86; 6.121–122; 8.8. Cf. Thuc. 1.20; 6.54–55.

16. Violations of natural boundaries may signify imperialism carried to excess (cf. 7.20, 35, 49, 54–57). For Croesus’ desire to conquer as a motive for his invasion of Cappadocia, see 1.73.1. Cf. Immerwahr (n.7 above), 84 and 293f.

17. Benardete, Seth, Herodotean Inquiries (The Hague 1969), 4f.Google Scholar; Lateiner, D., ‘Polarità: il principio della differenza complementare’, QS 22 (1985), 79–99, esp. 91fGoogle Scholar.

18. Immerwahr, H., ‘Aspects of Historical Causation in Herodotus’, TAPA 87 (1956), 241–80, esp. 247–51Google Scholar.

19. In Herodotus, temporal prōtos is mainly an explanatory marker of transition, which points to the beginning of a historical development (e.g. 1.6.2), but it has celebratory value when it identifies original cultural achievements and their inventors (e.g. 1.94.1). Sometimes, but not in this case, the two uses appear combined (1.163.1).

20. Flory (n.9 above), 420, regards it as ‘impulsive and misplaced’. Flory’s article has otherwise considerably helped my thinking on the subject of Arion.

21. The unity of the Histories on the basis of thematic analogy and patterning has been most convincingly demonstrated by Immerwahr (n.7 above), 79ff., and confirmed by, among others, Hunter (n.13 above), esp. 176–225. Contra Waters, K. H., Herodotus on Tyrants and Despots: A Study in Objectivity (Wiesbaden 1971 Google Scholar), passim.

22. Immerwahr (h.7 above), 35, calls the episode the first Greek logos of the Histories.

23. Cobet (n.4 above), 149.

24. Immerwahr (n.7 above), 154–61. Croesus is guilty of expansionism beyond the limits, caused by overconfidence (see n. 16 above), but he also pays for the crime by which the founder of his dynasty obtained the throne (see esp. 1.91).

25. Cf. Myres (n.9 above), who regards the Arion episode as an example of pedimental composition, where the shipmen’s crime and Arion’s rescue form the centrepiece.

26. See e.g. the different ways in which divine punishment strikes Candaules (1.7–13) and Cyrus (1.214.3–5) on the one hand, and Leotychides (6.72) and Cleomenes (6.75, 84.3) on the other.

27. For the Arion episode as anticipatory of the rescue of Croesus from the pyre, see Wood (n.3 above), 23f. According to Schwabl, H., ‘Herodot als Historiker and Erzähler’, Gymnasium 76 (1969), 253–72, esp. 259–61Google Scholar, both the singer and the dolphin in ‘Arion’ point to Apollo, so that there is an implicit Delphic link with the entire Croesus narrative. The legend, however, has also connections with Poseidon (Bowra [n.12 above], 133) and Dionysus (see n.30 below). Herodotus' silence on this point lets the story retain all its associations.

28. Evidence of divine presence in support of the Greeks is mentioned at 8.13, 65.2, 84.2, 94.2–3.

29. In itself the fact of the,rescue by a dolphin is not impossible and was not regarded so by the ancients. Bowra (n.12 above), 131.

30. That Arion in the dolphin legend is the heroic hypostasis of Dionysus has been urged by Deborah Lyons, whom I thank for making available to me her unpublished paper on ‘Arion and Dionysos Methymnaios’. See also Burkert, Walter, Homo Necans: The Anthropology of Ancient Greek Sacrificial and Ritual, tr. Peter Bing (Berkeley, Los Angeles and London, 1983), 199f Google Scholar.

31. Arion’s intrinsic worth distinguished him from the Homeric Chryses (Iliad 1. 9–42): Chryses (whose position as priest is symbolized by his ribboned staff, as Arion’s position as poet by his skeuē, ‘vestments’) is under divine protection specifically because of the external fact of his service to the god, and his prayer to Apollo, unlike Arion’s song, is an appeal, made on the basis of that claim.

32. The nomos orthios was a traditional religious hymn in honour of Apollo. See Stein, H., Herodotos I ⁶ (Berlin 1901 Google Scholar), ad. loc.

33. See the analysis of Herodotus’ description made by Flory (n.9 above), 412–14. Among other instances in Herodotus of a similar kind of heroism, Flory most appropriately cites (415f.) the episode of the Spartans practising gymnastics and combing their hair before the battle of Thermopylae (7.208).

34. In the Histories, intelligent actions among Barbarians are frequent (e.g. 1.184–87), and the Greeks appear foolish on occasion (1.60; 5.97), but on the whole sophiē in all its various manifestations (wisdom, shrewdness, scientific knowledge, technical expertise of any kind) is an especially Greek feature which counterbalances the material assets of the East. See Solon (1.29–33), Thales (1.74.2, 75.3–4), Democedes (3.125.1, 130), and Artemisia (7.99; 8.68–69, 87–88, 102–103), among others.

35. See especially 1.5.3–4: starting from Croesus, its explanatory beginning, the logos will take Herodotus to ‘cities great and small’, i.e., as Herodotus explains, objects of report that are remarkable on the basis of their greatness, past or present. For Herodotus’ use of the word logos or logoi in reference to his work, see Jacoby (n.4 above), 281f. and 327.

36. That Herodotus’ logos grows naturally according to an inherited tradition of oral narrative where association plays an important part has been maintained by Lang, Mabel L., Herodotean Narrative and Discourse (Cambridge, Mass., and London 1984), 1–17Google Scholar. However, Lang also observes that exceptionally the transition to the Arion story is not characteristic of the oral style, since it requires looking backwards (8 n.20). This observation confirms my argument that Herodotus has exploited the Periander connection in a very deliberate manner.

37. 1.30.4–5 (Tellus of Athens), 31 (Cleobis and Biton); 5.92 (tyranny in Corinth); 6.86 (Glaucus).

38. See e.g. 1.194.1, where Herodotus introduces the topic of the collapsible leather boats that the Assyrians use to take their merchandise down the Euphrates as a thōma megiston moi (‘a very great wonder in my opinion’). To Herodotus these primitive vessels are an important phenomenon, ‘second only to the city of Babylon itself,’ perhaps because they are an ingenious device by which the Assyrians exploit the river in a productive and legitimate way, in contrast with so many violations of rivers recorded throughout the Histories (see n.16 above).

39. Herodotus gives evidence of selectivity with regard to conventionally praiseworthy deeds for example at 5.72.4, and with regard to unreliable stories at 1.95.1.

40. Only an analysis of the individual cases can help to determine the significance of these landmarked phenomena as evidence, and the present discussion of the exemplary Arion episode only intends to set the premises for such an analysis. Many Herodotean ethnographical thōmata, especially monuments (e.g. 2.148.6, 155.3), seem to reveal only a conventional admiration for Oriental splendour (cf. Drews, Robers, The Greek Accounts of Eastern History [Cambridge, Mass., 1973], 49f.Google Scholar), but in the Histories all manifestations of greatness are placed in relation to other aspects of a culture as they appear reflected in customs, attitudes and historical actions.

41. I wish to thank A. John Graham and Donald Lateiner for reading the various drafts of this article and for offering many helpful suggestions.