¹ The main conclusion of this paper—that the mephitic chasm under the temple never existed, and that the real chasm is the Castalian gorge—has been definitely asserted by Prof.von Wilamowitz-Moellendorff, . Cf. Hermes xxxviii. 1904, p. 579Google Scholar. Arist. und Athen ii. p. 44, n. 17. He has, however, published no arguments, as far as I am aware, supporting his contention.

² There is no ancient authority for the description of the vapours as mephitic, but that is the invariable epithet applied to them in modern books. It does not seem to possess a very clear meaning. Servius ad Aen. vii. 84 says ‘Mephitis proprie est terrae putor qui de aquis nascitur sulfuratis et est in nemoribus gravior ex densitate silvarum.’ Modern writers seem mostly to connect it with volcanic fumes. The notion that any natural gas can create a prophetic excitement is totally erroneous and therefore it does not matter much what is understood by mephitic; cf. infra § viii. For the analogy with alcohol cf. Myers, Greek Oracles, p. 34 n.

³ Choephoroe 806 τὸ δὲ καλῶς κτάμενον. ὧ μέγα ναίων στόμιον. The scholiast notes ῾ὦ Αἴδη.᾿ Dr. Verrall is clearly right in refusing to accept κτίμενον the emendation of Bamberger, , and depart from the ancient tradition. A reference to his article in this Journal xiv. 8Google Scholar will shew that his decision has not been influenced by any prejudice against the existence of a chasm at Delphi. The word στόμιον is far more applicable to Hades than to Apollo, who would scarcely have been recognised by this appellation. The context moreover, in which the gods invoked are Zeus, the gods of the house and finally Hermes, who was especially connected with the dead, inevitably suggests that Hades and not Apollo is the god intended here.

⁴ See note at end of paper.

⁵ Eur., Phoen. 232Google Scholarἄντρα δράκοντος, ζάθєά τ'. I. T. 1245Google Scholarὅθι ποικιλόνωτος οἰνωπὸς δράκων σκιєρᾷ κατάχαλκος єὐφύλλῳ δάφνᾳ, γᾶς πєλωριον τέρας, ἄμφєπє μαντєῖον χθόνιον.

⁶ Clearohus, ap. Athen. 701Google Scholar c (F.H.G. ii. 318). The story is told in explanation of the phrase ἴε παιών J. H. Middleton was quite unjustified in identifying this σπήλαιον with the adytum; the story of the dragon-slaying as told in the Hom. Hymn is quite independent of the oracular chasm.

⁷ The full identification of the oracle with the chasm in this myth appears in Apollod., Bibl. i. 4. 1. 3Google Scholar. The dragon guarded the oracle and prevented Apollo from approaching the chasm.

⁸ ad Plut. 39. Cf. Origen, v. Cels. iii. c. 25, vii. c. 3Google Scholar (ed. Koetschau, Berlin, 1899). Chrysostom, Joh., Hom. xxix. ad i. Cor. c. 12, v. 1Google Scholar, Migne, , Patr. Gr. lxi. 242Google Scholar. The origin of the scholium was noted by eighteenth century scholars (cf. Dindorf ad loc.) yet it is repeatedly quoted as an authority, notably by Middleton, , J.H.S. ix. 305Google Scholar, who gives a false reference. For a different explanation of the story cf. Vergil, NordenAen. vi. 1. 77Google Scholar.

⁹ The words used by Diodorus προσβλέψασαν (of the goats) and κατιδὁντα (of the shepherd) and the absence of any reference to a πνεῦμα ἐνθουσιαστικὁν seem to imply that he looked upon the inspiration as visual in origin. Clearly antiquity was very doubtful about the mephitic vapours; but these minor inconsistencies in such poor evidence are not of much account. The matters which are explained by the aetiological legend quoted by Diodorus are: α. The ‘oracle from goats’ practised at Delphi, i.e. the preliminary rite known to us from Plutarch (de Def. Or. 46, 49, 51). β. The attribution of the oracle to Γῆ. γ. The invention of the Tripod. δ. The virginal dress of the Pythia.

¹⁰ Varro, de Ling. Lat. vii. 17Google Scholar, cf. J.H.S. ix. p. 294 n. I am unable to agree that the ‘quod vocant Delphis in aede ad latus est quiddam ut thesauri specie quod Graeci vocant ὀμφαλόν quem Pythonis aiunt esse tumulos’ of Varro means ‘what in the temple at Delphi is called the χάσμα (the oracular cleft in the rock).’ His words suggest rather that there was an omphalos tomb as Rohde, , Psyche ³ i. 132Google Scholar, and MissHarrison, , J.H.S. xix. 225Google Scholar, conclude to have been the ease. To this question I hope to return on another occasion.

¹¹ Plutarch took part as priest in the dedication of a statue to Hadrian, cf. C.I.G. i. 1713. Pomtow, , Jahrb. 1889, 551Google Scholarsqq.

¹² c. 42 καὶ γὰρ ἐνταῦθα τὴν περὶ τὸν τόπον δύ ναμιν ἐμφανῆ γενέσθαι πρῶτον ἱστοροῦσιν, νομέως τινὸς ἐμπεσόντος κατά τινα τὑχην, εἶτα φωνὰς ἀνα φέροντος ἐνθουσιώσεις κ.τ.λ. . . .οἱ δὲ λογιώτστοι Δελφῶν καὶ τοῦνομα τοῦ ἀνθρώπου διαμνημονεύ οντες Κορήταν λέγουσιν

c. 46 ὁ γὰρ Κορἡτας ἐκεῖνος, ὄν Δελφοὶ λέγουσιν πρῶτον ἐμπεσόντα τῆς περὶ τὸν τόπον δυνάμεως αἴσθησιν παρασχεῖν

¹³ Ammonius the speaker who makes the second reference to Koretas dismisses the story in the words εἴ γε δὴ τοῦτο μὴ μῦθὁς ἐστι μηδὲ πλάσμα κενόν, ὡς ἔγωγ᾿ ἡγοῦμαι. The importance of the choice of speaker will be noted below.

¹⁴ Middleton, J.H.S. ix. 304Google Scholar asserts that ‘Plutarch in many passages speaks of the Pythia descending to the vault or ἄντρον.’ This is presumably a misprint for ἄδυτον since in the two references given there is no mention of an ἄντρον, nor is there as far as I am aware in any writing of Plutarch's. There certainly is not in the treatises on the oracles. But if a misprint it is a very misleading one.

¹⁵ Hy. Hom. ad Apoll. 443. In l. 523 Allen and Sykes read δεῖξε δ᾿ ἄγων ἄδυτον ζἁθεον καὶ πίονα νηόν from an inferior group of MSS. for αὐτοῦ δάπεδον. Here, if there is any distinction between ἄδυτον and νηός it would seem that the former is the wider term. The identification of the adytum, as opposed to the temple, with the building of Trophonius and Agamedes spoken of by Steph. Byz. s.v. Δελφοί is not very convincing and the second interpretation put forward by Allen and Sykes (p. 103, ad. 1. 296) is preferable.

¹⁶ Cf. Eur., Andr. 1105Google Scholar. Ion 79, 102, 422, etc.

¹⁷ Cf. Pind, Ol. vii. 32Google Scholarὁ χρυσοκόμας εὐώδεος ἐξ ἀδύτου . . . εἶπεon which the schol. remarks ἀδύτου ἤγουν τοῦ ναοῦ τῆς εὐωδίας πεπληρωένου So Aristonous hymn l. 14 φρικώεντος ἐξ ἀδὑτου of Apollo. These are of course mere phrases. How they are to be understood appears from the story of Aristodicus at Branchidae (Hdt. i. 159) who went round the outside of the temple destroying the birds and their nests when λέγεται φωνὴν ἐκ τοῦ ἀδύτου γενέσθαι κ.τ.λ.. Similarly Euripides, , I. T. 973Google Scholar says that when Orestes—on the occasion of his second appeal to Apollo—lay πρόσθεν ἀδύτων ἐκταθείς, ἐντεῦθεν αὐδὴν τριποδος ἐκ χρυσοῦ λακὼν Φοῖβός μ᾿ ἔπεμψε. In none of these passages is there any notion of an inner chamber. Euripides also uses the word ἄδυτα very vaguely in I. T. 1257 (text corrupt) and Andr. 1035. A similar vague use of the word τῶν γνὀφων πνεύμασι πλέοντες. τὰ ὰπὸ γνόφων appears in Ion 739, where it = the whole πνεύματα

¹⁸ The first person met with on leaving the temple is described in various ways. Hdt. vi. 34 ὁ δὲ ἀπὸ γῆς καὶ συστροφῆς ἀέρος γνοφίας. αὖραι γὰρ καὶ αὐταὶ καὶ ῥύσεις Magnesian Inscr. Kern. 17. l. 28 ἀέρων τυγχάνουσιν οὖσαι, καὶ οὐκ ἀλόγως ἄνεμοι and l. 38 καλοῦνται, ὅτε ἤ ἀπὸ λιμνῶν ἤ ποταμῶν φέρονται

¹⁹ Plutarch, (de E ap. Delph. 2)Google Scholar says that no woman may approach the oracle. Euripides seems to have known of no such restriction.

²⁰ Similar commands to leave the temple occur in other oracles, cf. Ael., var. hist. iii. 43Google Scholar. ὁμοῖοι δὲ τούτων εἰσὶ καὶ οί ἀπόγειοι Galen, , Protr. c. 9Google Scholar, καταβάλλειν For the λινουργοί of Ael., v. h. iii. 44Google Scholar Simplicius, ad Epictet. encheir. xxxii. has λιμουργοί. An oracle quoted by Zosimus i. 57 as delivered by Apollo Sarpedonios to the people of Palmyra has ΓΒ ΓΓ

²¹ Hdt. v. 72. He also speaks of a Γνώμῃ on the Acropolis (viii. 53) into which Athenians fled when the Persianshad made their way thither by the secret ascent. It is the same as the adytum in this passage; cf. Dörpfeld, , Ath. Mitth. xii. p. 27Google Scholar. In ix. 116 there is mention of the Γράμματι Βουλῆς, Γερουσίας of Protesilaos at Elaieus, which may or may not have been underground. If it was, Artayktes had a queer idea of comfort, but his sacrilege would have been as great if the adytum merely means the temple, cf. Plut., v. Demetr. 24Google Scholar. For οἱ γὰρ Αἰγαῖοι φιλοτιμίαν ἀνόητον ἐπανελό as applied to Branchidae cf. supra n. 17. There were certainly adyta underground, cf. Frazer, , Paus. vol. iii. p. 15Google Scholar (from whose instances v. 1. 5 and x. 33. 11 should be deducted as uncertain), but the theory that they owed their name μενοι πρὸς ὑμᾶς τὸ περὶ τὰς ἀπογραφὰς ἐξαμαρτἀ to their origin from caves (cf. Smith, W. Robertson, Rel. of Sem. ² p. 200)Google Scholar receives no confirmation from the use of that word in Herodotus. He uses it ii. 141, 143, 169, 176 of the main chamber of Egyptian temples, vi. 134 of the adytum of Demeter at Paros, v. 77 of a Western Hall on the Acropolis at Athens. Of these only the one at Paros seems to have been a secret chamber, but nothing suggests that it was subterranean.

²² de Pyth. Or. c. 6 νοντες κ.τ.λ. ΓΠ ΓΠΒ c. 22 Γνώμῃ Πρυτάνεως, Γνώμῃ Πρυτάνεως Βουλῆς (γνώμῃ) c. 28 ὄταν δ᾿ ἐκεῖ κατέλθῃ καὶ γένηται παρὰ τῷ θεῷ de Def. Or. 8 δυσὶν ἐχρῶντο προφήτισιν ἐν μέρει καθιεμέναις c. 51 κατέβη εἰς τὸ μαντεῖον, ὤς φασιν, ἄκουσα καὶ ἀπρόθυμος As far as I know no word such as ἀναβαίνειν is ever used for her departure from the adytum as Rayet, asserts Milet et le Golfe Latmique, part 2, c. 2, p. 60Google Scholar. Études, p. 156.

²³ Vit. Tim. c. 8. εἰς Δελφοὺς πορευθεὶς ἔθυσε τῷ θεῷ (cf. Hdt. vii. 140 etc.) καὶ καταβαίνον τος εἰς τὸ μαντεῖον αὐτοῦ γίγνεται σημεῖον. ἐκ γὰρ τῶν κρεμαμένων ἀναθημάτων ταινία τις ἀπορρυεῖσα καὶ φερομένη στεφάνους ἔχουσα καὶ Νίκας ἐμπεποικιλμένας περιέπεσε τῇ κεφαλῇ τοῦ Τιμ Cf. Hermann, , Gr. Alt. ii. p. 259Google Scholar, n. 16. Frazer, , Paus. vol. v. p. 353Google Scholar.

²⁴ De Pyth. Or. c. 26. οὐ γὰρ ὁ δεῖνα μὰ Δία κατέβαινε περὶ ὠνῆς ἀνδραπόδσυ χρησόμενος οὐδ᾿ ὁ δεῖνα περὶ ἐργασίας The use of the word without an object would suggest that the technical meaning of the term is connected with the consultant rather than with the Pythia, and this is borne out by the passages from earlier authors quoted below.

²⁵ M. Homolle concludes that the present temple was exactly similar to that built by the Alcmeonids and destroyed in the fourth century. Cf. B.C.H. xx. 1896, p. 654.

²⁶ Cf. Frazer, , Paus. v. 316Google Scholar, passages there quoted. For the inscription and the masonmarks, Homolle, , C.R. Acad. Inscr. xxiii. p. 335Google Scholar n. Even those who hold that the omphalos was originally a tomb do not dispute its ultimate removal to the temple.

²⁷ For the meaning of the word οἶκος cf. Fraenkel, and Furtwaengler, in Rh. Mus. lvii. 1902, 152, 252, 543Google Scholar. Delamare, , Rev. Ét. Gr. xvi. 1903, p. 160Google Scholar. The use of the word here agrees with none of the meanings suggested.

²⁸ Tektonik ² ii. § 62. 4. Rayet's theory that it was a vestibule similar to that discovered at Branchidae receives no support from the name of the latter, cf. Pontremoli, and Haussoullier, , Didymes, p. 92Google Scholar. But for this theory, of. infra, nn. 36, 39.

²⁹ The whole notion is an extension of a phrase of Pindar, , Ol. vii. 32Google Scholar, supra n. 17, and therefore the word adytum is used while it is elsewhere avoided by Plutarch, cf. supra n. 23. Incidentally we may notice that the sweet-smelling adytum is very far from the mephitic chasm or the cave whose lower depths were fatal. Of course in Pindar the epithet is quite commonplace.

³⁰ Cf. Wilamowitz, , Isyllos p. 11Google Scholar, I.G. iv. 950. 29 οὐδέ κε Θεσσαλίας ἐν Τρίκκηι πειραθείης είς ἄδυ τον καταβὰς Ασκληπίου κ.τ.λ. We do not know enough about Trikka to say with certainty that it possessed no underground adytum, but it is very unlikely. For Epidaurus cf. Dittenb., Syll. 803Google ScholarKavvadias, , Fouilles, 12Google Scholar. I. G. iv. 952. 112. There certainly the sleeping place was not subterranean. The use of καταβάς in connexion with incubation may be due to analogy with the regular words κατακοιμᾶν κατακεῖσθαι of the consultants. (Professor Ernest Gardner suggests to me that the inscription should read ἄβατον and not ἄδυτον I have no means of checking Kavvadias' reading.)

³¹ Hymn. Hom. ad Apoll. 443. The commentators do not seem to have remarked the irregular use of the aorist active in place of the middle.

³² In quoting this oracle Dio Chrys. xxxvii. p. 103 reads εἰσαφικάνει and when the same line recurs in an oracle ap. Certamen Homeri et Hesiodi, p. 17 ἀμφιπολεύει is introduced from Hymn xxiv. 1, ῾ Ιστίη ἤ τε ἄνακτος ᾿ Απόλλωνος ἐκάτοιο Πυθοῖ ἐν ἠγαθέῃ ἰερὸν δόμον ἀμφιπιλεύει where ἀμφιπολεύει is thoroughly applicable to Hestia who dwelt within, but scarcely suitable to Hesiod. The two last feet had to be filled up. Another oracle, Anth. Pal. xiv. 77, varies the phrase ὄλβιος οὔτος ἀνὴρ ὄς νῦν κατὰ λάϊνον οὐδὸι Φοίβου ᾿ Απόλλωνος χρηστήριον εἰσανα βαίνει So ibid. xiv. 100, Menelaos and Paris are said ἐμὸν δόμον εἰσανέβητε (schol. A. Il. v. 64). The fact that the form which occurs in Hdt. is the nearest to Epic usage seems to prove it genuine.

³³ Il. ix. 704, Od. viii. 80, Hy. Hom. 296. It occurs with the tripods in or. ap. Aelian V. H. iii. 43, Anth. Pal. xiv. 77. For a possible meaning cf. infra, note 39.

³⁴ It is not impossible that Plutarch's choice of the word was partly influenced by its technical sense = in certamen descendere, cf. L. and S. s.v. 1. 4. So too καθίημι is used and, in late Greek, κατέρχομαι For various senses of the word and of ἀναβαίνειν cf. Dörpfeld, u. Reisch, Theater, p. 189Google Scholar.

³⁵ The two classical instances are variants of the same story, applied to different personages— Philomelos (Diod. Sic. xvi. 27) and Alexander (Plut., Alex. 14)Google Scholar. In both the excuse given for a refusal is that the day was not the regular one—at least in Diod. that seems to be the case, though the text is corrupt. Her reluctance in the case cited above is due to the unfavourable omens. Only Roman authors—Valerius, Lucan—make it personal. Cf. Gardner, and Jevons, Manual p. 265Google Scholar for the traditional view of the Pythia's frenzy, and Middleton, J.H.S. ix. 304Google Scholar for the argument thence to the intoxicating air.

³⁶ Cf. Frazer, , Paus. v. 336, 353Google Scholar. In 1894 Homolle, M. thought he had found traces of the ‘sombre et souterrain adyton’ C.R. xxii. 587Google Scholar presumably where ‘le dallage est coupé vers le milieu du monument par une dépression large et profonde’ (B.C.H. xviii. 177. C.R. xxii. 302, Arch. Anz. 1895, p. 4). But in 1897 he writes B.C.H. xxi. 273 that he has not found ‘l'antre de la Pythie.’ The reconstruction of the temple at Branchidae to which the remains at Delphi were at first compared itself owes too much to a fancied theory of the temple of Delphi to be quoted as evidence. If it does represent the tradition it would appear that the descent into the cella or adytum was a real one, but, as the arguments above would shew, that descent was made by prophet and consultant alike; and the adytum, if lower in level than the prodomos, was not subterranean or secret, but was hypaethral and the chief hall of the temple. Cf. Rayet, Études, p. 153Google Scholar.

³⁷ B.C.H. xx. 643, cf. Frazer, Paus. v. 399Google Scholar, for Mycenaean remains and tombs generally.

³⁸ Homolle quoted by Studniczka, Hermes, xxxvii. 263Google Scholar.

³⁹ Cf. Frazer, Paus. vol. v. p. 333Google Scholar. This account is later in date than the provisional report of Homolle, M., B.C.H. xviii. 1894, 177Google Scholar, C.R. xxii. p. 302, which is contradicted in one important detail (the existence of an interior colonnade) by the report of the next year, C.R. xxiii. 329, and conflicts with the suggested restoration recently published (Fouilles de Delphes ii. Pl. 6). The existence of an opisthodomos is attested by inscriptions; C.R. xxiii. 335, and seems to disprove the suggestions of Rayet, cf. supra, n. 36.

⁴⁰ The belief that oracular chasms were volcanic is very hard to kill. The compiler of the Guide Joanne has discovered a mephitic chasm in a fancied cave of Apollo Pythios at Thera. Certainly volcanic soils and mephitic vapours abound in that region, but the limestone mountain of Meso Vouno, where the cave is situated, is perhaps the only place in the island where they are impossible. The association of Apollo Pythios with caves is largely a fallacy produced by this belief in ‘mephitic vapours.’

⁴¹ For the geology of Delphi see Dr. Alfred Philippson in Pauly-Wissowa Real-Encycl. s.v. Delphoi. The statements made above I owe to the kindness of Dr. Philippson himself who with the greatest courtesy has sent me privately answers to all my questions. He himself considers the story of a mephitic chasm ‘im Ganzen Priester-Betrug gewesen zu sein.’

⁴² Castalia is the one spring which is invariably mentioned by ancient authors as the prominent Delphic feature, e.g. Pind., Pyth. iv. 163Google Scholarμεμάντευμαι δ᾿ ἐπὶ Κασταλίᾳ It was certainly used for ceremonial washings, Eur., Ion 94, 146Google Scholar, Phoen. 222, Aristonous, Hymn. v. viGoogle Scholar, Horace, Od. iii. 4. 61Google Scholar, but the unanimity of the evidence to its supreme importance as the spring par excellence of Delphi is the only good ground we possess for connecting it with the water drunk by the priestess. Paus. x. 24. 7 states that she drank the water of Cassotis.

⁴³ Cf. Hymn. Hom. ad Apoll. 300 and Sykes and Allen ad loc. Euripides and others as we have seen speak only of the cave, but in late authors, e.g. Ovid, , Metamorph, iii. 14Google Scholar, Origen vii. c. 3, the cave is called Castalium antram, τὸ τῆς Κασταλίας στόμιον Their testimony is of little importance but the Homeric Hymn is explicit, and seems to prove that the Python is no gravesnake as MissHarrison, would make it, J.H.S. xix. 205Google Scholarsqq., but the ordinary guardian of the spring, cf. Frazer, , Paus. v. 44Google Scholar. This spring can hardly have been Cassotis if as Dr. Frazer says that source is fed by a conduit, p. 355, nor could it have been in the sanctuary of Earth since springs do not gush out of the schist. The water there is also brought by a conduit (Frazer, op. cit. p. 286).

⁴⁴ Whether the oracle at Oropus was the original Amphiareum or not does not seem to matter here. If it was transferred here from elsewhere the place can only have been chosen because of some pre-existing and similar rites. But cf. Wilamowitz-Moellendorff, , Hermes xxi, p. 104Google Scholar n. 2.

⁴⁵ We hear of a cave at Klaros, , Tac., Ann. ii. 54Google Scholar. It has perhaps been found situated in a chasm similar to those mentioned above, cf. Schuchart, , A.M. xi. 432Google Scholar. It must have been entirely distinct from the temple and not unlike such a Sibyl's grotto as was artificially constructed and adorned in the 2nd century A.D. at Erythrae, cf. Buresch, , A.M. xvii. 16Google Scholar. On the other hand the supposed oracular cave at Ptoion for which there is no literary evidence is a humble construction of stone very near the foundations of the temple (cf. Frazer, , Paus. v. 100)Google Scholar. All these caves contain water either from a spring or brought by a conduit. The temple at Branchidae (cf. supra, n. 36) dates from the third century, and further excavations must be awaited before the account of Rayet and Thomas is accepted.