¹ The term ‘proof’ can be used in a rigorous philosophical sense to denote the process of reaching conclusions from indemonstrable prior premises, via valid inferences, i.e. as an equivalent to ‘demonstrate’. It is important to note, though, that it is often used loosely – i.e. for informal notions of proof: e.g. a conclusion drawn from evidence or experience which seems in the circumstances to be as decisive as one can expect (e.g. proving guilt in a law court ‘beyond reasonable doubt’). A claim to ‘have proof’ can mean that there is evidence pointing clearly to a certain conclusion, without being absolutely demonstrable. Tekmerion hovers around this latter complex of meanings: often denoting evidence which is being adduced as proof for a certain conclusion, or as part of proof (or proofs) which the author believes is decisive in the circumstances. Thus one wants to translate tekmerion sometimes as evidence (leading towards a certain conclusion), or sometimes as proof in the informal sense, since it is used as the critical evidence or argument which leads to a conclusion the writer regards as certain.

² I will not discuss here the precise forms of argumentation used or the apodeixis and epideixis (which are important and involve wider issues).

³ Cf. the most thorough examination of Herodotus' intellectual links, Nestle, W., Herodots Verhältnis auf Philosophie und Sophistik (1908)Google Scholar.

⁴ Lloyd, A., Herodotus Book II, Introduction (1975), esp. ch. 4Google Scholar; Gould, J., Herodotus (1989)Google Scholar; cf. Dihle, A., ‘Herodot und die Sophistik’, Philologus 106 (1962) 207–20CrossRefGoogle Scholar, on some sophistic links, and see now Fowler, R. L., ‘Herodotus and his contemporaries’, JHS 116 (1996) 62–87, esp. 86–7CrossRefGoogle Scholar for Herodotus as a σοφός. Cf. Lateiner, D., ‘The empirical element in the methods of the early Greek medical writers and Herodotus: a shared epistemological response’, Antichthon 20 (1986) 1–20CrossRefGoogle ScholarPubMed for links with medicine.

⁵ See Fornara, C. W., ‘Evidence for the date of Herodotus' publication’, JHS 91 (1971) 25–34CrossRefGoogle Scholar and ‘Herodotus' knowledge of the Archidamian War’, Hermes 109 (1981) 149–56Google Scholar; but he probably downdates too far.

⁶ For the problem of Herodotus' autopsy see esp. Schepens, G., L' ‘Autopsie’ dans la méthode des historiens grecs du Ve siècle avant J.-C. (1980)Google Scholar; Lateiner 1986 (n.4); J. Gould, Herodotus ch.1–2 generally; and articles by Armayor, O. K., ‘Did Herodotus ever go to the Black Sea?’, HSCP 82 (1978) 45–62Google Scholar; ‘Sesostris and Herodotus' autopsy of Thrace, Colchis, inland Asia Minor and the Levant’, J. Amer.Research Center in Egypt 15 (1980) 59–71Google Scholar; Herodotus' autopsy of the Fayoum (1985); with works cited in following note.

⁷ Fehling, D., Herodotus and his ‘sources’. Citation, invention and narrative art (1989)Google Scholar (translation and revision of Die Quellenangaben bei Herodot 1971); but cf. Marincola's highly apposite remarks (that they are not specifically source citations anyway), Arethusa 20 (1987) 26–32 and 126Google Scholar; also Erbse, H., Studien zum Verständnis Herodots (1992) 76 n.6CrossRefGoogle Scholar, and his ‘Fiktion und Wahrheit im Werke Herodots’, GGN 1991 131–50Google ScholarPubMed. For further discussion on his ethnography Redfield, J., ‘Herodotus the tourist’, Class, Phil. 80 (1985) 97–118CrossRefGoogle Scholar; Pritchett, , The liar school of Herodotus (1993)Google Scholar; Gould, J., Herodotus (1987) ch.5Google Scholar; Konstan, D., ‘Persians, Greeks and empire’, Arethusa 20 (1987) 59–73Google Scholar.

⁸ Müller, D., ‘Herodot – Vater des Empirismus? Mensch und Erkenntnis im Denken Herodots’, in Kurz, G. et al. , Gnomosyne: menschliches Denken und Handeln in der frühgriechischen Literatur. Festschrift W. Marg (1981) 299–318Google Scholar.

⁹ Lateiner (n.4); and more generally The historical method of Herodotus (1989) esp. ch.9 ‘Fact and explanation’.

¹⁰ Hartog, F., The mirror of Herodotus. The representation of the other in the writing of history (transl. Lloyd, J. 1988Google Scholar; orig. Le Miroir d'Hérodote: essai sur la représentation de l'autre 1980); Konstan, (n.7); Cartledge, P., The Greeks (1993)Google Scholar.

¹¹ Note the stimulating general treatment of Calame, C., Mythe et histoire dans l'antiquité grecque (1996)Google Scholar, an important critique of the myth–history and mythos–logos divisions; and S. Humphreys, Serious stories: narrative, proof and persuasion in classical Greek prose (forthcoming), on wider issues of authority and persuasion in the period.

¹² See Gomme's note (A historical commentary on Thucydides) on 1.1.3 and 1.20.1 for tekmerion: rather rigidly, he insists (on 1.20.1) ‘tekmerion is not evidence but the inference drawn from the evidence’; but his confident distinction between marturia and tekmeria is not borne out by other examples in Thucydides, nor by rhetorical usage: see Hornblower, , Thucydides (1987), 100–7Google Scholar.

¹³ Other Thucydidean uses of tekmeria in connection with his own methods: 2.15.4 (acropolis and polis); 2.50.2 (plague: proof for effect on birds); 1.20.1 (evidence); 1.21.1 (, and ). Homer also offers evidence: 1.3.3 (); 1.9.4; 3.104.6 (for Delos). See also n.46 below for rhetorical contexts.

¹⁴ Hornblower, , Commentary on 1.1.3 (p.7)Google Scholar – true if one means straight explicit declaration of his methods in a separate section rather than embedded in a specific context. Cf. also his Thucydides, 100–7, with the interesting suggestion that Thucydides' vocabulary of evidence is related to that of the lawcourts (particularly rich in bks. 1–2) but not to Aristotle's or later technical distinctions between semeion and tekmerion (cf. 1.132.1 and 5).

¹⁵ Solmsen, F., Intellectual experiments of the Greek enlightenment (1975) 228Google Scholar; he implies that Thucydides in his great argument about the Peisistratids (6.54–9) talks of tekmeria and semeia: he may use them, but seems almost deliberately to avoid this language.

¹⁶ Cf. Manetti, G., Theories of the sign in classical antiquity (1993; Ital., ed. 1987)Google Scholar ch.2 (though cf. 45 on Hdt. 2.33); cf. 37: ‘the debt which serious historiography, inaugurated by Thucydides at the end of the fifth century B.C., owes to Hippocratic techne has also been pointed out.’ (referring to Vegetti: see n. 18). Manetti mostly uses Herodotus for inference in divination; contrast Lloyd, G. E. R., Polarity and analogy (1966) 341–5Google Scholar on Herodotus' use of analogy and inference (in comparison with Hippocratics).

¹⁷ σταθμάομαι and δηλόω are also relevant; note also ἀπόδειξις, ἐπίδειξις and cognates which I hope to discuss elsewhere. Semeion also comes to mean ‘sign’ in the sense of ‘evidence’ in philosophical and medical writers, but in Herodotus it retains its primary sense of ‘mark’.

¹⁸ See esp. Lloyd, G. E. R., Magic, reason and experience (1979) ch.3 ‘Development of empirical research’Google Scholar; ch.2, ‘Dialectic and demonstration’; also Revolutions of wisdom (1987)Google Scholar; Manetti (n.16); Vegetti, (ed.) Opere di Ippocrate (1976)Google Scholar; Lonie, Iain M., The Hippocratic treatises ‘On generation’, ‘On the nature of the child', Diseases IV’ (1981) esp. 72 ffGoogle Scholar.; Diller, H., ‘ΟΨΙΣΑΔΗΛΩΝΤΑ ΦΑΙΝΟΜΕΝΑ’, Hermes 67 (1932), 14–42Google Scholar.

¹⁹ Müller (n.8).

²⁰ Lateiner (n.4). On ancient medicine's argument in favour of observable evidence and experience, the evidence of the senses, epitomizes Lateiner's definition of the empirical.

²¹ Lateiner (n.4) 11; cf. his Historical method, 192–3.

²² ‘For the Greek historian, an empirical method was more easily applied [than for a natural scientist], because in the realm of human events, the truth sought and the reality investigated were not so dependent on technological sophistication’ (Lateiner (n.4) 8); though cf. provisos at 3.

²³ For which, in addition to works cited in n. 18, see Burnyeat, M., ‘The origins of non-deductive inference’, in Barnes, J., Brunschwig, J., Burnyeat, M., Schofield, M., Science and speculation (1982) 193–238Google Scholar; Wenskus, Otta, ‘Vergleich und Beweis im Hippokratischen Corpus’, in Formes de pensée …, Actes IVe 1981 (1983) 393–406Google Scholar. Manetti ((n.16) 40) finds it surprising that already in Sacr. dis. tekmerion appears with the sense of proof, ‘reliable sign’.

²⁴ Fehling (n.7) 17–21, 131–2.

²⁵ The two other occurences of tekmerion are: 7.16 γ 2, in Artabanus' speech to Xerxes on the interpretation of the dream: ; and 7.234.1, Xerxes to Demaratus on Demaratus' qualities of character: . The Pelasgian argument of 1.57.1–2 is discussed below.

²⁶ As also pointed out of Thucydides (see n. 12).

²⁷ Other refs. to marturia etc. in Hdt.: 2.22.2 (his tightly argued passage on the Nile flooding); 8.55 (olive tree and the sea are marturia for the contention of Poseidon and Athena). For the absolute sense of bearing witness cf. from numerous cases: 5.24.3 (Darius' own words) and 8.94.4 (Corinthians' claim about their conduct at Salamis is witnessed to by the rest of Greece – a quasi-forensic context). For the evidence and counter-evidence of Sybarites and Crotonians, see Kukofka, Dirk-Achim, ‘Das der Sybariten (Herodot 5.43–46)’, Hermes 119 (1991) 374–80Google Scholar.

²⁸ Cf. also Eum. 797–9: Athena tells the Furies how the case was equal, how Zeus provided μαϱτύϱια, as did the oracle of Apollo.

²⁹ Tekmeria in tragedy, along with cognates (tekmar, etc.): tekmerion occurs in Aeschylus (9 times and once in V.; tekmairomai once only, in PV), in Euripides (15 times; tekmairomai 5 times), and least often in Sophocles (4 times; the verb twice). Tekmerion seems to be used in a very similar sense to tekmar (‘sign’), i.e. as a sign, evidence (often qualified with ‘clear’, ‘sure’), often regarded as decisive, but comparatively simple from which to infer the situation: e.g. Agam. 352 (tekmeria for Clytemnestra's announcement that Troy had fallen – to be read with de Jong's, I. note, ‘ in Soph. El. 774’, Mnemosyne 47 (1994) 679–81)CrossRefGoogle Scholar; Soph., El. 1109Google Scholar (where the are for Orestes' death and refer to the physical remains); or Aesch., Suppl. 54Google Scholar (the Danaids' that they are Argives involve Io's tale). In Euripides tekmeria are more obviously decisive evidence or proof, often in recognition scenes (Ion 237, 329, 349, etc.), often contrasted with doxa or some form of uncertainty (HF 714; Hipp. 925, Med. 517), and particularly for the problem of judging human character (Med. 517, Hipp. 280, 925; fr. 60). These latter are perhaps echoes of the lawcourts.

³⁰ Eurip. fr.811 N (Phoenix) ; fr.574 N (Oinomaos) . Some other Euripidean references may be found in previous note.

³¹ Heraclitus makes general, aphoristic statements about the senses: DK 22, B107 ‘the eyes are bad witnesses for men, and ears for those who have barbarian souls’; B101a, B34. Empedocles makes open appeal to evidence once, DK 31, B21. Xenophanes used the evidence of fossils (DK 21, A33: Hippolytus), but unfortunately we do not know how he chose to do so.

³² B1: .

³³ Note also Xenophanes, DK 21 B18 which indicates a process of showing, searching and discovering.

³⁴ DK 87, fr.44, B col. 1, line 10: unfortunately fragmentary. Cf. Finley, J.H., Three essays on Thucydides (1967) 74CrossRefGoogle Scholar.

³⁵ There is not, of course, universal agreement on the dating of these essays, but the following are generally agreed to be early: Airs, Nature of man, Breaths, Art, Generation/Nature of the child; On ancient medicine, Sacred disease. Epidemics 1 and 3. The sophistic style of Art and Breaths has been discussed most recently by Jouanna, Budé text, Hippocrate V 1; the rhetorical nature of these and other early essays is stressed by Lloyd, G.E.R., Magic, 86–98Google ScholarPubMed, and Revolutions, 91–100, 114–23, and by Jouanna, Hippocrate, and his ‘Rhétorique et médecine dans la Collection hippocratique’, REG 97 (1984) 26–44CrossRefGoogle Scholar; Kudlien, F., ‘Dialektik und Medizin in der Antike’, Medizinhistorisches Journal 9 (1974) 187–200Google Scholar.

³⁶ Discussed by Diller ( n.18) 17–19: not strictly an analogy but an experiment using signs and inference. For experiments in medicine see esp.: Bourgey, L., Observation et expérience chez les médecins de la collection hippocratique (1953)Google Scholar; Lloyd, G. E. R., Magic, ch.3, esp. 146–60Google ScholarPubMed.

³⁷ Cf. also other important passages in which the language of proof and deduction is used: Airs ch.3 init. (CMG I 1 (2).26.22) ; ch. 10 (CMG I 1(2).46.17) , plus discussion of semeia for whether the year will be healthy (CMG I 1(2).46.19); ch.ll (CMG I 1(2).52.15) ; and ch.24 (CMG I 1(2)82.14–15) . Cf. also ch.9 (CMG I 1(2).46.4).

³⁸ On ancient medicine ch. 17 (I 612.10–11 L; 17.2 Jouanna), imagining an opposing objection about the effect of heat: ‘but I consider that this is my strongest evidence that men are not feverish only through heat (); ch. 18 (I 612.19 L; 18.1 J) ‘This is clear from the following pieces of evidence (); and summing up an argument, ch.8 fin. (I 588.2 L; 8.3 J) ‘These are all proofs (ταῦτα ) that the art of medicine, if research is continued on the same lines, can all be discovered’.

³⁹ Cf. On the art ch.5 (5.3 Jouanna; VI 8.3–5 L) ‘and it is surely strong proof of the existence of the art, that it both exists and is powerful’ () – because even people who do not believe in it recover with its help. A little later, mistakes are also evidence for an art: (ch.5.5 J; VI 8.16–17 L). Various symptoms such as the roughness of the voice, ‘furnish medicine with the means of inferring what conditions these symptoms indicate’ (Jones): (ch. 13 Jones; 12.2 J; VI 24.6 L). Note also TEτεϰμαϱεῖται and τεϰμαίϱεται at ch. 12.4 J (VI 24.11, 15 L).

⁴⁰ There are other relevant expressions, e.g. which occurs three times at Nat. man ch.7 (CMG I 1(3). 182.12, 19; 184.6).

⁴¹ For the overlap of medicine and philosophy: G. E. R. Lloyd, Magic (passim); Jones, W. H. S., Philosophy and medicine in ancient Greece, Bull. of History of Medicine Suppl. 8, (1946)Google Scholar; Longrigg, J., Greek rational medicine from Alcmaeon to the Alexandrians (1993)CrossRefGoogle Scholar; Jouanna, J., Hippocrate p. 366–403Google Scholar.

⁴² Eg. Regimen in acute diseases ch.39.3 Joly (II 306.13 L) , which seems to denote some form of ‘proof’, but with connotations of teaching; also at 39.1 Joly (II 304.10 L). Note also ch.20.2 Joly (II 268.5 L) θεωϱέων, ‘keeping a sharp eye for the signs already noted’ (with reference to writing, unusual in the essays discussed above); πϱοστεϰμαϱτέα, ‘one should also take account of, ch. 38.1 Joly (II 302.11 L); ἐς τέϰμαϱσιν, ‘for the interpretation of symptoms’, ch.1.1 and 2.1 Joly (II 224.8 L bis), (Thucydides uses this form in a speech, 2.87.1, ‘ground for inference’); also the more familiar marturia at ch.39.1 Joly (II 304. 8 L ϰϱέσσον μαϱτύϱιον); and semeia (‘symptoms’), ch.44.2 (II 318.6 L), ch.48.2 (II 330.4–5 L) and, less technical, ch. 6.1 Joly (II 236.2 L). For historion see Sauge, A., De l'épopée á l'histoire. Fondement de la notion d'historié (1992) 202–10Google Scholar. There may be difficulties in determining neologisms, given the nature of our evidence, but that most of these examples seem to have relatively narrow life-span and provenance would reinforce the impression that they are new coinages.

⁴³ Note also Breaths, ch. 12.2 Jouanna (that breaths are the source of disease) and ch. 12.3. In On the sacred disease: ch. 5 (VI 364.20–1 L) tekmeria appear in the midst of the argument that the disease is not more sacred than any other, ; ch.7 (VI 368.6 L) on the breath in the veins, and the proof, τεϰμήϱιονδέ. ὃταν… See note 45 for Epid. 1 and 3; similarly in Prognostic ch. 2.13 (II 116. 2 L), semeia, and throughout; also cf. ch. 24.71ff. (II 188.2–3 L) (which may show neatly the distinction between the two roots); and ch. 25.11f. (II 188.14–15 L) in conclusion, εὖ (that in every year and country, bad signs indicate something bad, and vice versa).

⁴⁴ For the first person in Herodotus see Dewald, C., ‘Narrative surface and authorial voice in Herodotus' Histories’ in Herodotus and the invention of history, Arethusa 20 (1987) 147–70Google Scholar; J. Marincola, ‘Herodotean narrative and the narrator's presence’, ibid. 121–37; and cf. J. Jouanna, ‘Rhétorique et médecine (n.35); and Lloyd, G. E. R., Revolutions of Wisdom 56–70Google Scholar.

⁴⁵ For semeia, e.g. Epid. 1, ch.10, lines 6–8 Jones (); ch.25.14–15; and cf. Epid. 3, ch.10, line 5 Jones, ἐσήμαινεν.

⁴⁶ Rhetorical contexts in Thucydides: 1.34.5 (Corcyraean speech); 1.73.5 (Athenian speech); 3.66 (Theban speech); 2.39.2 (Periclean speech); 6.28.2, accusations (indirect) against Alcibiades; 2.87.1 τέϰμαϱσις in a speech of exhortation. At 1.132.4, there is a hint of the forensic. Also 4.123.2 (Mendaioi). Of his own methods see n. 13.

⁴⁷ Cf. Lysias 16.11 .

⁴⁸ The early development of rhetoric is notoriously difficult to understand and the evidence frustratingly fragmentary. See works cited in n.51.

⁴⁹ The principle of recording what has been said (even if he does not believe it) is most openly expressed at 2.123.1; also 7.152.3.

⁵⁰ The text at 1.56–8 has caused problems, but they have no bearing on the argument presented here about Herodotus' language. Cf. McNeal, R. A., ‘How did the Pelasgians become Hellenes? Hdt. 1.56–58’, Ill.Class.Stud. 10 (1995) 11–21Google Scholar: but he neglects the fact that Herodotus' arguments are about language.

⁵¹ For some work on proof and argumentation in rhetoric: Schupp, F., ‘Zur Geschichte der Beweistopik in der älten griechischen Gerichtsrede’, Wien.Stud. 45 (1926) 17–28 and 173–85Google Scholar, at 26 on the category of proof ‘which refers generally to the topos’ (e.g. Antiphon Tetral. A δ 8); Radermacher, L., Artium scriptores (1951) 209, 214–15Google Scholar; Due, Bodil, Antiphon. A study in argumentation (1980)Google Scholar; cf. also Kennedy, G. A., Art of persuasion (1963)Google Scholar, and A new history of classical rhetoric (1994).