¹ I adopt the words ‘routes’ in preference to ‘roads’ lest I should convey a wrong impression. Traces of ancient made roads in the Peloponnese are extremely rare; those which occur on the routes here investigated will be noticed in their proper places. The large majority of ancient, as of modern, routes in the peninsula were mere mountain-tracks, identifiable only by the objects in their neighbourhood or by topographical considerations.

² viii. 44. 1—3.

³ Of these (1) ascends left of Auemodhoúri, and descends right of Marmariá; (2) ascends right of Anemodhoúri, descends left of Pápari; (3) ascends by Skortsinoú, descends at Koutri-boúkhi or Páperi. The last of these, which passes far south of the summit, will be further described in connexion with one of the Spartan routes (pp. 48–9).

⁴ These heights are obtained by averaging the results of a number of observations made with an aneroid.

⁵ Paus. viii. 36. 5.

⁶ Paus. viii. 3. 1, 2.

⁷ Herod. ix. 11; Plut., Arist. 10Google Scholar; Thuc. v. 64.

⁸ Leake, , Peloponnesiaca, pp. 247Google Scholar, sq.

⁹ Bursian, , Geographic von Griechenland, vol. ii. 227 (and note 3).Google Scholar

¹⁰ Vol. ii. pp. 318, sq.

¹¹ Loc. cit.

¹² The only remains there ale those of a very small building, or enclosure, of loose stones. The peasants call it a chapel of St. Elias; but it is so rude and shapeless that it may possibly have never been anything but a sheep-pen.

¹³ Plut., Cleom. 4Google Scholar.

¹⁴ See pp. 48–49.

¹⁵ The breadth of the flutes in the fragment of column is approximately 2¾ inches. The length of the metope and triglyph combined is 2 ft. 6 in., of the triglyph alone 10½ in. The height of the metope and triglyph is 1 ft. 5 in. The architectural forms are those of a rather late period.

¹⁶ Possibly from Dholianá in the hills southeast of Tegea, where there are both ancient and modern quarries.

¹⁷ See Pans. viii. 3. 1.

¹⁸ Ibid.

¹⁹ Id. 38. 1.

²⁹ Id. 39. 3—5.

²¹ Herod, ix. 11; Plut., Arist. 10.Google Scholar

²² Thuc. v. 64.

²³ Paus. viii. 27. 3.—Oresthasium has frequently been confused, or consciously identified, with ‘Orestia’—an old name for the part of Megalopolis which lay south of the Hclisson. That Orestia was half of Megalopolis is expressly stated by Steph. Ryz. s.v. That it was the southern half is clear (1) from Thuc. iv. 134, where the expression doubtles refers to the place afterwards known as Ladokeia, a southern suburb of Megalopolis (Paus. viii. 44. 1), (2) from Paus. viii. 34. 1—4, where a series of monuments commemorating the story of Orestes are mentioned on the road to Messene. That it was an old name for this region appears from the fact that the passage of Thucydides quoted above refers to a period long before the foundation of Megalopolis.

Orestia derived its name from Orestes (Steph. Byz. s.v. and we may infer the same from Paus. loc. cit.), Oresthasimn from Oresthens (Paus. viii. 3. 1 ; Steph. Byz. s.v. ); but both were occasionally called ‘Oresteinm,’—the former in Eur., Orest. 1647Google Scholar (cf. id.Electra 1273—5), the latter in Herod. ix. 11, Plut., Arist. 10Google Scholar, Paus. viii. 3. 2 (cf. also Thuc. v. 64, ). The two places are merged in the article ‘Oresthasium’ in Smith's Dict, of Gr. and Roman Geog.; and the same mistake (for I feel sure it is a mistake) led Boblaye (Recherches pp. 172, 173), Bursian (ii. 227, and note 3), and apparently Curtius, (Peloponnesos i. 316)Google Scholar, to place Oresthasium west of Tsimbaroú, i.e. in the Megalopolitan plain.

²⁴ Cp. Gell, , Itinerary of the Morea, p. 97.Google Scholar

²⁵ Thuc. iv. 134.

²⁶ Polyb. ii. 51, 55; Plut., Arat. 36, 37.Google Scholar

²⁷ I have, however, never been able to see any traces of the ruined walls mentioned near this spot by Boblaye, (Recherches, p. 173)Google Scholar. On the other hand, Boblaye's remark, that this site will not suit the Athenaeum rests on a confusion of this Athenaeum with (Plut., Cleom. 4Google Scholar), which must have been a totally different place (v. infra, p. 39).

²³ viii. 44. 3—8.

²⁹ Appendix A.

³⁰ The order was Dorie. The flutes, in the extant fragments of columns, range from 3½ in. to 4¼ in. in width. There are also fragments of triglyphs. Both I and 1 clamps were used.

So far as one can judge from the remains of its foundations, the temple was peristyle, its external measurements being approximately 95 ft. × 40 ft., and the external measurements of the cella about 74 ft. × 22½ ft. But these are very rough measurements, the best that can be obtained without a complete clearance of the site and the removal of the ruins of a chapel of Ag. Demos which has been superposed.

The temple on Ag. Elias bas not, I believe, been hitherto noticed by archaeologists. In Baedeker's Guide the summit of the hill has been marked erroneously as the site of Asea; and the remains described as existing upon it in the text of the same handbook (p. 299) are really those of the acropolis of Asea, to which I have already referred in the text.

³¹ These are principally fragments of Dorie columns, the width of the flutes ranging from rather more than 3 in. to rather less than 4 in. A grave objection to identifying this temple with that of Athena Soteira and Poseidon lies in the interpretation of the words which should properly mean ‘at the top of the mountain.’ But on the whole this seems to me less objectionable than the forced interpretation of which is the alternative to it.

It is practically certain that the real summit of Krávari, which I have visited more than once, is not the site of a temple.

³² Bobl., Recherches, pp. 143, 173Google Scholar; Ross, , Reisen im Peloponnes, p. 61Google Scholar (Ross's apparent dissent from the French explorers here arises from a misunderstanding of their map; he and they really held precisely the same view); Curt. i. 262; Burs. ii. 217; etc.

³³ Vol. i. Pl. III.

³⁴ It is certainly strange that the boundary should have been down in the plain, so that both slopes of Krávari were included in the Megalopolitan territory. But the fact is quite clear from Pausanias' description, and is admitted on all hands. The boundary between the Tegean and Argive territories, near Hysiae, was very similarly situated. See Paus. viii. 54. 7, and p. 79 of the present paper.

³⁵ I believe it is never entirely dry, even in the height of summer, but I do not mean to assert this too positively.

³⁶ This identification was first, I believe, made by Ross, and his view has been generally accepted. The French explorers had previously supposed the big hill of Ag. Elias just east of Kaparéli, or the lower slopes of it, to represent Kresium (Expéd. Scient, de Morée, Atlas, Pl. IV.), and were followed by Leake in his Peloponnesiaca (special map of the Mantinice and Tegealis at end of volume). The principal objections to this view are: (1) the height of the hill—perhaps 800 ft. or 1,000 ft. above the plain—to which the words seem inapplicable; (2) its great distance to the right of the direct route from the Choma to Tegea. Had this been Kresium, Kresium would have been mentioned in connexion with the route from Tegea to Sparta rather than in connexion with that from Megalopolis to Tegea.

For the word applied to a very small, but conspicuous, elevation in a plain, cf. Paus. viii. 12.7, where it is applied to the hillock on which stood Old Mautineia.

³⁷ Expéd. Scient, de Moréc, Atlas, Pl. IV.; copied by Leake in his special plan at the end of Peloponnesiaca, Curtius (vol. i. Pl. III.), and Bursian (vol. ii. Pl. VI.).

³⁸ For the alternative forms v. Paus. viii. 45. 1, id. 47.1; Steph. Byz. s.v.

³⁹ This hill may be easily distinguished by a group of three oak-trees which crowns it.

⁴⁰ Paus. iii. 20. 8—21.3.

⁴¹ Where Tyndareus sacrificed a horse on the occasion of the oath taken by Helen's suitors.

⁴² Here follows a story of Penelope's departure from Sparta with her husband Odysseus.

⁴³ There are various forms of the name: (at least the region is Strabo, p. 343), and probably (Paus. viii. 27.4) Except in translation, where the name must be given literally, I prefer the form ‘Belmina’ to ‘Belemina’ chiefly for the sake of consistency, since one must occasionally speak of the ‘Belminatis’ or ‘Belminatid territory’ (cf. Polyb. ii. 54 and Strabo p. 343), and ‘Beleminatid’ is intolerable.

⁴⁴ Paus. viii. 35. 3.

⁴⁵ Travels in the Morca, iii. 20.

⁴⁶ Leake was told there were ‘Hellenic ruins’ there. But the peasants, who are still unanimous in asserting the existence of remains, failed altogether to show me any traces of them. There is however sufficient evidence that minor antiquities, notably some small bronze figures, have occasionally been found near the spot.

⁴⁷ Boblaye, , Recherches, pp. 75Google Scholar, sq.; Curtius, ii. 256, sq.; Bursian, ii. 113. Boblaye speaks of the ‘ruines très étendues vues par M. Vietty sur le plateau, au sud de Pétrina.’ This is, I suspect, the only evidence for the remains which rests on actual observation, and even this evidence is only reported at second-hand. I cannot help thinking that M. Vietty was taken in either by the appearance, at a distance, of this region, which is studded with white rocks protruding from the soil, or by the traces of some loose stone walls—probably of cottages or sheep-pens—which may be of any date. In rebuilding the chapel of Ag. Theódhoros, which is in this region, some walls of stones, tile, and mortar, were discovered; and also a very small marble Doric capital, of such bad workmanship that it must be of very late, probably Byzantine, date. I suspect there was a small monastery here.

⁴⁸ Cleom. 4.

⁴⁹ Probably to be translated ‘entrance to’ or ‘pass into Lakonia.’

⁵⁰ Polyb. iv. 37, 60, 81.

⁵¹ Id. ii. 46.

⁵² For the successive changes of ownership of Belmina, the Belminatid territory, and the Athenaeum, the following passages should be consulted in the order in which I give them:

(1) Paus. viii. 27. 4 (if Blenina = Belmina), cf. ib. 35. 4; (2) Plut., Cleom. 4Google Scholar and Pol. ii. 46; (3) Pol. ii. 54; (4) Pol. iv. 37, 60, and 81; (5) Livy xxxviii. 34; (6) Paus. iii. 21. 3, viii. 35. 4.

⁵³ The single exception is the prominent hill whose summit lies almost exactly in a line between the villages of Petrina and Grikoń, from each of which it is, however, separated by a deep stream-bed. This hill overhangs the eastern (or modern) track to Megalopolis, and commands a good view of the western track (the one commonly used in ancient times, and now used by travellers to Leondári). It also commands a magnificent view of the Megalopolitan plain as far north as Karýtaena, and conceals this plain from Khelmós. Were there any traces of a fortress on the summit, one might naturally identify it with the ‘Athenaeum’; but there are none, and a rocky summit like this could hardly have been fortified without retaining traces of the fact.

⁵⁴ Unless, indeed, ‘Athenaeum’ was specially a Megalopolitan name for the fortress of Belmina.

⁵⁵ Bursian (ii. 113, note 3) expressly rejects the identification of Khelmós with the Athenaeum on the ground that the position of Khelmós will not suit the other Athenaeum, which he does not distinguish from this one.

⁵⁶ Supposing, that is, that the ancient track kept close to the river at this point, coinciding perhaps (as far as the ‘bridge of Kopános’) with the track to Tegea. But it seems more probable that (like the modern one) it made a short cut by passing to the left of the low rocky hill described in the text.

⁵⁷ Baedeker, who (p. 280) mentions polygonal remains ‘on a hill on the opposite bank,’ probably refers, though inaccurately, to this site.

⁵⁸ If it represents any of the objects noted by Pausanias, it must be this. The words suit it well. But, if so, the statue of Aidos must have been at some point a good deal less than ‘thirty stades’ from Sparta. In the first plate to Ross's Reisen im Peloponnes, ‘Grab des Ladas’ is marked near this spot, but without any apparent reference to extant remains.

⁵⁹ Travels in the Morea, iii. 13.

⁶⁰ Id. 15.

⁶¹ Bursian, ii. 115. See also Baedeker, , p. 280, and the Guide Joanne, p. 281.Google Scholar

⁶² They are very convenient for lighting fires in. Hence the name. I have myself seen traces of fires there. From the name a story has arisen that they were the cooking-places of the workmen employed in making the aqueduct from the ‘Vivári’ to Sparta.

⁶³ As suggested in Baedeker, p. 280.

⁶⁴ The ‘Hellenikó’ of the Guide-books.

⁶⁵ Cf. Bursian, ii. 114, sq.

⁶⁶ Leake, , Travels in the Morca, iii. 13Google Scholar, sq.; and subsequent writers

⁶⁷ It may be roughly estimated as varying, in different places, from 50 to 200 yards in width.

⁶⁸ See above, p. 43.

⁶⁹ The following story, with variations, told by the rustics in this connexion.—A certain Greek princess having two suitors, set them each a labour to porform. One was to bring water from the ‘Vivári’ to Sparta, the other to build the fortress of Mistrá; and the one who first accomplished his task was to win her hand. Unfortunately the two tasks were completed simultaneously; and the princess, unable either to satisfy the claims of both her suitors or to choose between them, took poison and died.

⁷⁰ This is probably the of Plut., Agis, 8Google Scholar.

⁷¹ Possibly both; but at any rate the theory attaches to the one in the plain. It is probably erroneous, for the water (especially that of the lower spring) is too clean and too cold to have run so short a distance underground. The ancient theory was that the water of the ‘Táka’ rose again at Frangóvryso, but the comparative levels of the two plains makes this impossible. V. Appendix A, p. 68.

⁷² Pol. iv. 81.

⁷³ Livy xxxv. 27.

⁷⁴ The former by Leake, , Travels in the Morca, iii. 18Google Scholar, sq. and Bursian ii. 114; the latter by Leake, , Peloponnesiaca (which represents his later views), p. 350Google Scholar, and Curtius ii. 258.

⁷⁵ Supposing that in Paus. viii. 27. 4

⁷⁶ E.g. Pol. ii. 54.

⁷⁷ An old well, completely filled up, has also been recently discovered at the east end of the acropolis, just below the crown of the hill. It is about six feet in diameter, and is cut chiefly in soft rock. The proprietor has excavated it to a depth of some 30 feet in hope of finding treasure.

⁷⁸ Paus. viii. 35. 3, 4.

⁷⁹ These identifications are necessitated (1) by the distance from Megalopolis, as given in the present passage, (2) by Paus. viii. 44, 4, and 54, 3, where we are told that the sources of the Alpheius in the Megalopolitan territory derive their water from the plain of Asea (Frangóvryso)—a theory which could not have arisen had the Alpheius been any river other than the one I have described. On the relation between the waters of the Asean plain and the Alpheius, see Appendix A.

⁸⁰ Peloponnesiaca, p. 237.

⁸¹ Itinerary of the Morea, p. 213.

⁸² The audience must have taken place late in the day, and the march have been a rapid one.

⁸³ viii. 3. 2.

⁸⁴ P. 28 sqq.

⁸⁵ Paus. viii. 27. 3.

⁸⁶ P. 28.

⁸⁷ The extreme danger of this as a military route is well illustrated by a story which I heard from a very old Turk at Longaniko. Ibrahim Pasha (he says), marching from Sparta to Tripolitsá, was led by a treacherous guide into this gorge, whereas he ought to have gone by Skortsinoú. He immediately found himself shot at from the overhanging hills on both sides, and, after losing many of his men, eventually succeeded in effecting his escape viâ Kalteziá

⁸⁸ The highest point is in that part of the route which coincides with (2).

⁸⁹ Probably the name itself refers to this.

⁹⁰ viii. 27. 3.

⁹¹ Hell. vi. 5. 12, 20, 21.

⁹² Travels in the Marea, iii. 31–33.

⁹³ The ‘Paleó-khora’ mentioned by Leake (admittedly on hearsay evidence) is almost certainly the site of a deserted village, not (as he supposed) of an ancient town. It is probably identical with the ‘Vili. r^e’ (= Village ruiné?) of the French map, where I have seen traces of rude walls, but nothing which suggests antiquity. On the other hand, by a small spring on the way from Barbítsa to Frangóvryso, about ten minutes' walk from the former, and just before one reaches the chapel of Ag. Giánnes, there exists a piece of wall, of rather carefully fitted polygonal work, nearly fifty feet in length, preserved in parts to a height of over four feet. This wall, which I excavated (for scarcely a trace of it was visible above ground), appears to be almost isolated, and I was quite unable to discover to what sort of building it belonged.

⁹⁴ It has the bulging echinus which generally characterizes an archaic order; but the profile is in other respects so peculiar as to suggest that it is slovenly work of late date. It has been hollowed into a rude trough, and lies just outside the door of the priest's house.

⁹⁵ Hell. vi. 5. 12.

⁹⁶ See Pl. II.

⁹⁷ viii. 53. 11.

⁹⁸ Id. 54. 1, sqq.

⁹⁹ Here follows an account of the Alpheius crossing the Adriatic and reappearing at Syracuse.

¹⁰⁰ When I was there (June, 1893), literally nothing; the small driblet which flowed from it being diverted so as to water a bed of onions or garlic close by. But there was at that time not so much is a drop of water either in the main stream-bed or in the one which here joins it from the direction of the khan of Bakoúros.

It is a pity that the spring of Kryávrysi, which used to be treasured by travellers as supplying the last drink of fresh water on the way from Tripoliná to Sparta till one reached the khan of Vonrliá, is now utterly neglected and its drinking basin destroyed. The khan of Kryávrysi, too, is deserted. The disappearance of this (like that of many another old institution in Greece) is due to the construction of the carriage-road. The latter skirts the hills a great deal higher up, missing the old spring and khan.

¹⁰¹ With this difference, that the track from Tripolitsá passes west of Kamári and does not enter the gorge till it reaches a point just below Mavríki; while, starting from Pialí or any other of the villages on the site of Tegea, one naturally goes by the gorge all the way.

¹⁰² Peloponnesiaca, pp. 114, sq. and map. The theory does not appear in Travels in the Morea, which (unlike Peloponnesiaca was based wholly on personal observation.

¹⁰³ Bulletin de Corr. Hell. vol. xvi. (1892), p. 534 and Pl. XIII.

¹⁰⁴ See Appendix A, pp. 68–9.

¹⁰⁵ Peloponnesiaca, p. 113.

¹⁰⁶ ii. 38. 7.

¹⁰⁷ From the ‘khan of Bakoúros’ to beyond the ‘khan of Kokkinóloutsa’ the mule track and new carriage road practically coincide; then a divergence begins, the mule track eventually joining the Arákhova route, by the river-side, at the now ruined ‘khan of Krevatás,’ while the carriage road climbs the hills west of it. A little further on, just before one reaches the ‘khan of Vourliá,’ all three routes coincide for a short distance. See Pl. I.

¹⁰⁸ Thuc. v. 55. 3; Xen., Hell. vi. 5. 25–27Google Scholar; Livy xxxiv. 26, xxxv. 27.

¹⁰⁹ This appears (1) from Thuc. v. 55. 3, where the Spartans make their (un-successfully) at Karyae; (2) from Livy xxxiv. 26 compared with ib. xxxv. 27. In the former passage T. Quinctius pitches his camp at Karyae, and there waits before entering Lakonian territory; while in the latter, which describes events which occurred only three years later, Philopoemen is said to have encamped at Karyae in Lakonian territory. Whether it had changed hands in the interval, or not (and there is no indication that it had), it was clearly a border town. But (3) that it did change hands more than once on other occasions is clear from Paus. viii. 45. 1, which makes it originally Tegean—id. iv. 16. 9, which makes it Spartan (in the time of the second Messenian war)—Xen., Hell. vi. 5Google Scholar. 25, where Karyae, after the battle of Leuktra, secedes to the enemies of Sparta; and id. vii. 1. 28, where it is re-taken by the Spartans and severely punished for its treachery.

¹¹⁰ Paus. iii. 10. 7.

¹¹¹ Leake, who originally supposed (Travels in the Marea, ii. 510) that the Argos-Sparta route crossed the Parnon range by a pass south of the summit, viâ Kastánitsa, afterwards (Peloponnesiaca, pp. 298, 339) abandoned that view in favour of the more correct identification of the French topographers. No one who has tried the two routes can be in any doubt about the matter. The pass by Kastánitsa is far higher, steeper, and rockier, than the other. It is a very trying route for a single pedestrian, or for a loaded mule, and would be almost intolerable for an army; while the Arákhova route is, as Greek mountain-routes go, a remarkably simple and straightforward one.

Further, we know from Polybius (ii. 65) that a part of the route of Antigonus from Argos to Sparta, viâ Sellasia, was and that the battle of Sellasia itself was fought actually in and about the river-bed. Now all this applies perfectly to the Arákhova route, but not to that by Kastánitsa; for the river of Tsintsina, and its junction with that from Agrianós, lie at the bottom of ravines so deep and difficult that the track goes far above them—there is no room for a path, much less for a battle, in the river-bed.

Lastly, Karyae, which we know to have been situated near the Tegean and Spartan frontier, would certainly never have been mentioned by Pausanias (iii. 10. 7) in connexion with the Argos-Sparta route at all, if that route had passed as far south as the river of Tsintsina.

I believe the Hermae which marked the common boundary of the Argive, Tegean, and Spartan territories, and which Pausanias (ii. 38. 7) describes as [sc. ], to have been at the top of the pass; not (as Jochmus, suggests in the Journ. Roy. Geog. Soc. 1857, p. 43)Google Scholar a little west of Meligoú. The three large heaps of stones which may still be seen close to the path, within about 100 yards of the top of the pass, at a spot called doubtless commemorate some murder or massacre; but it may very possibly have been a murder or massacre of quite recent date. Stories which connect them with the battle of 300 Spartans against 300 Argives (Baedeker, p. 263), or with Herakles and Hippokoon, (Journ. Roy. Geog. Soc. 1857, p. 42Google Scholar), are merely the guesses of half-educated priests or schoolmasters with a smattering of ancient Greek; not, as Jochmus imagined, traditions ‘perpetuated from the earliest ages of Grecian antiquity.’

¹¹² Reisen im Peloponnes, p. 175.

¹¹³ ii. 261.

¹¹⁴ ii. 118.

¹¹⁵ Partly owing to the new road taking a different line from the old one. The ‘P. K.’ must not be confused with the ‘pýrgo’ (πύργο), a conspicuous mediaeval ruin much nearer Arákhova.

¹¹⁶ This site appears to be unknown to the topographical handbooks from Leake's time downwards; but is, if I mistake not, the same which Lieut. -Gen. Jochmus independently identified with Karyae, (Journ. Roy. Geog. Soc. 1857, p. 49 and map)Google Scholar. My attention was first called to Joohmus' paper by Mr. J. G. Frazer. It is a valuable contribution to the topography of this region.

¹¹⁷ xvi. 37.

¹¹⁸ iii. 10. 6.

¹¹⁹ Hell. vi. 5. 27.

¹²⁰ iii. 10.6–11.1. For proof that this route, like that from Tegea, went viâ A rákhova, see note 111.

¹²¹ Hell. vi. 5. 27.

¹²² The remains are of much too permanent a character to be those of one of the camps in which Kleomenes entrenched himself before the battle of Sellasia, as suggested by Boblaye (p. 74); nor is it possible to adapt such a theory to Polybius' account of the battle.

¹²³ Peloponnesiaca, pp. 343, sq.

¹²⁴ Livy xxxv. 27 sqq.

¹²⁵ Id. xxxiv. 26 sqq.

¹²⁶ P. 64.

¹²⁷ Livy xxxv. 27.

¹²⁸ Polyb. ii. 65 sqq.; Plut., Cleom. 27, 28Google Scholar; id.Philop. 6; Paus. viii. 49. 5, 6.

¹²⁹ The only exception, so far as I know, is Leake; who, though he modified his original views after the appearance of the French publication, never fell in with the French identification of the site [v. note 131). For what may be considered the orthodox explanation of the battle, based on the French identification, v. Ross, , Reisen im Peloponnes, pp. 181Google Scholarsqq., and Map.

¹³⁰ The so-called ‘Eva’ is a high hill, standing far away from the river-bed which is sup posed to have been the scene of the battle; and the so-called ‘Olympus’ is not an individual hill, but part of the skirts of the mountain which over-hangs the river from its junction with the river of Vréstena downwards.

¹³¹ Leake at different times proposed two different sites for the battle, both of which had the advantage of being at the junction of genuine rivers, but both of which topographers have rightly discarded. His first theory (Travels in the Morea, ii. 526 sqq.), which placed the battle a little above the monastery of Ag. Saránda (‘the Forty Saints’), at the junction of the rivers of Tsíntsina and Agrianós, was based on a mistaken view as to the route by which Antigonus had marched from Argolis—a mistake which has been sufficiently discussed above (note 111). His second theory (Peloponnesiaca, 341–349) placed it a little below the monastery —where the river of Arákhova joins the combined streams from Agrianós and Tsíntsina— and resulted from an attempt to reconcile his former view (that Sellasia itself was near the monastery) with the undoubted fact that the route of Antigonus was not (as he had formerly supposed) viâ Kastánitsa, but viâ Arákhova. But this revision of his theory was made many years after he had visited the spot, and is quite untenable. For (1) the route to Sparta must certainly have passed, like the modern track as well as the carriage road, west of the hill of Ag. Konstantínos, and not through the difficult and dangerous gorge along which the river flows; (2) even were it otherwise, it would have been madness on the part of Kleomenes to have left Ag. Konstantínos undefended, and to have opposed Antigonus at the exit, instead of the entrance, of the gorge; (3) the proposed site, though not so completely shut in as the one which Leake had formerly selected, is too confined to admit of any battle in which a large number of troops, including cavalry, were employed.

¹³² Hell. vi. 5. 22 sqq.

¹³³ xv. 63 sqq.

¹³⁴ Xenophon distinctly says Mantineia: while from Diodorus it appears that the meeting took place just after a defeat of the Spartans by the Arkadians at Orchomenus, about five hours' journey north of that town.

¹³⁵ (Diod. xv. 64). In the Teubner edition of 1867 the word is bracketed, and in that of 1893 omitted altogether, as being unintelligible—in opposition to all the MSS. Is it possible that it meant ‘straight,’ the commonest meaning of the word (as well as of the adverb ) in modern Greek? Even, however, if it be omitted, the fact that this route to Sellasia, alone of the four, is left wholly undescribed, seems to imply that it was the ordinary one.

¹³⁶ iii. 10. 8.

¹³⁷ Travela in the Morea, iii. 28.

¹³⁸ Id. iii. 30.

¹³⁹ In the French map a ‘P. K.’ (Palaeókastro, παλαιό-καστρο) has been marked, not indeed on Ag. Khristóphoros, but on another part of the same range of hills, at a point bearing approximately south-east from Ag. Khristóphoros and north-east from Kolínaes. If I am right in identifying the hill thus indicated with that now called Ag. Elias, the evidences of antiquity there are (at present at any rate) quite inadequate. Nor is there any other hill in the neighbourhood which the villagers of Kolínaes can point to as bearing any traces of an ancient site.

¹⁴⁰ ‘Arvaníto-Kerasiá’ and ‘Vlákho-Kerasiá’ = ‘Albanian’ and ‘Wallachian’ Kerasiá respectively. Both villages now contain a mixed population.

¹⁴¹ They consist of (1) remains of a wall of hewn masonry, slightly polygonal in character, forming part of a large, partially rock-cut, building; (2) other rock-cuttings; (3) a profusion of pottery, some of it with black glaze.

¹⁴² On the occasion of the expedition of Agesilaus against Mantineia, only a year before, we hear of the various Arkadian contingents meeting at Asea (Xen., Hell. vi. 5.Google Scholar 11).

¹⁴³ xv. 64.

¹⁴⁴ Paus. ii. 38. 7.

¹⁴⁵ Travels in the Morea, iii. 29.

¹⁴⁶ The evidence on this point is not quite clear. See Paus. viii. 27. 4 and 35. 4.

¹⁴⁷ Hell. vi. 5. 24.

¹⁴⁸ Paus. ii. 38. 7.

¹⁴⁹ ‘Pyrrhi Castra’ has sometimes been identified with the ‘Πύῤῥου χάραξ’ of Polyb. v. 19. But this must be a mistake; for to reach the latter Philip κ α τ έ β η from Amyklae in the direction of the sea. The name appears to have been a common one. A ‘Castra Pyrrhi’ in Illyria is mentioned by Livy (xxxii. 13).

¹⁵⁰ Journ. Roy. Geog. Soc., 1857, p. 52.

¹⁵¹ Livy xxxv. 27.

¹⁵² In Leake's, sketch (Morea ii. 530Google Scholar) these rivers are marked as flowing from Vérria and Tzitzina (= Tsíntsina) respectively; but a comparison of this sketch with the French map or with Pl. I. will show that they are more correctly described as the rivers of Tsíntsina (and incidentally Vérria) and Adrianós.

¹⁵³ Pharae (Livy xxxv. 30) has been placed by Leake and others, owing partly to the resemblance of names, at Vérria, a very small village lying about half-an-hour's walk north-east of Basará (Leake, , Pelop. p. 345Google Scholar; cf. p. 53 of Jochmus', paper in the Journ. Roy. Geog. Soc. for 1857Google Scholar). The identification is little better than conjecture; but there are undoubtedly traces of wheel-ruts in the rocks between Basará and Vérria; and this tends to show that at or near Vérria there was an ancient town; for the road, of which they are evidence, must have been a local one, the route this way, viâ Kastánitsa, to the Thyreatid plain being altogether impassable to anything on wheels. If Pharae = Vérria, then the narrow track described by Livy, on which the battle took place, was probably on the road from Pharae to Sparta, not on that from Barbosthenes to Sparta. Philopoemen, it will be remembered, was on his way from Barbosthenes to Pyrrhi Castra, not to Sparta directly. The direct Sparta-Barbosthenes road (Livy xxxv. 30) may have followed the lower course of the river of Arákhova, between Mount Vréstena and the series of hills on which stand the remains of Sellasia and the villages of Voutiáni, Theológos, etc. But since the river here flows in a ravine so deep that the path, to get along at all, is obliged to skirt the hills at a considerable height above its left bank, I do not suppose, with Jochmus, that this was the ‘via patentior’ by which Philopoemen descended to the Eurotas. It seems to me more likely that he retraced his steps for a considerable distance, then descended to the valley of the river of Arákhova, and struck the ordinary Tegea-Sparta route at some point below the khan of Krevatás.

Leake, (Pelop. p. 344Google Scholarsqq. and map) identified Mount Barbosthenes with the hill of Ag. Konstantínos above the khan of Vourliá — the hill surmounted by the remains which are now-a-days generally supposed to be those of Sellasia. Sufficient reasons for the identification with Sellasia have been given already. But even leaving Sellasia out of account there are several considerations which forbid us to identify the hill with Mt. Barbosthenes. (1) It is on the direct route from Tegea to Sparta, and Barbosthenes was not. (2) It is not by any means ten Roman miles from Sparta. (3) ‘The defile in which stands the khan of Vourliá, and through which the modern road [i.e. the track which crosses the Eurotas by the ‘bridge of Kopanos’] from Tripolitsá to Mistrá descends to the Eurotas,'—in which Leake's views make it necessary to place the battle,—is not by any means such a defile as Livy describes. It is steep indeed in parts, but nowhere difficult, and its total length is very much less than the five miles which Livy gives as the length of Philopoemen's column alone. (4) There is no ‘via patentior’ to the Eurotas.

¹⁵⁴ Strabo, p. 275.

Id. p. 343.

¹⁵⁵ That the ‘Alpheius’ of Paus. viii. 54. 1 is the modern Sarandapotamós is generally admitted. See pp. 53–4.

¹⁵⁶ This mythical element appears in both Pausanias and Strabo, but is most prominent in Strabo, p. 275, where the superstition about the chaplets is narrated.

¹⁵⁷ Shown both in Pl. I. and in Pl. II.

¹⁵⁸ Peloponnesiaca, p. 114, and map.

¹⁵⁹ Bulletin de Corr. Hell. vol. xvi. (1892), p. 534 and Pl. XIII.

¹⁶⁰ Revue de Géographie, 1892, pp. 342 sq.

¹⁶¹ There must of course be holes in the rock beneath, which we do not see.

¹⁶² Of the two principal holes I find the following details in my notebook:—

(1) 7 Nov. 1891; ‘a’ almost completely choked, ‘b’ open, the stream flowing down it in a sort of waterfall to a depth of (say) 15 ft., and filtering through rubbish at the bottom.

(2) 28 Nov. 1891; ‘a’ open; no water getting so far as ‘b,’ but some filtering through the ground between the two.

(3) 14 June, 1892; water running down both ‘a’ and ‘b,’ principally the latter.

(4) 10 Nov. 1892; water just trickling down ‘a,’ none reaching ‘b.’

Any water which does not find its way down the katavothras runs beyond them to the gorge, as stated in the text.

¹⁶³ I once found one of them dry; but one at least (they tell me) runs throughout the summer.

¹⁶⁴ I have seen a considerable volume of water at the turbid springs when only a small quantity was trickling down the more obvious katavothras by Marmariá (10 Nov. 1892); but there was plenty of water in the stream in the Frangóvryso valley, and it gradually disappeared, before reaching the visible katavothras, by silent soakage. In fact, the stream bed, like some (probably many) others in Greece, appears not to be water-tight; it is, in fact, riddled with minute katavothras.

¹⁶⁵ See Pl. I. This is by far the most important spring in the Belminatid region, and one of the principal sources of the Eurotas.

¹⁶⁶ In the outer wall of the Athenaeum, the place where the path to Skortsinoú begins is known by the name ‘Πόρταις’ (‘the gates’), v. Fig. 6. The name may indicate an ancient tradition; but it is quite as likely that the theory of a gateway has arisen from this being the point at which the fortress is entered at the present day.

In the lower wall of Sellasia it may be conjectured that there was an ancient entrance in the west wall, a hundred yards or more from its southernmost point. The nature of the ground makes this a very convenient place to enter the ancient circuit, and for a few paces there are here no traces of the wall.

¹⁶⁷ For an account of the walls of Megalopolis, and a comparison of them with other fortification walls, v. Excavations at Megalopolis 1890—1891, pp. 108 sqq.

¹⁶⁸ It should be stated that at one point in the outer wall, where it forms part of the upper enclosure,—at a point where it is preserved to a considerable height,—several courses of smaller stones appear above the larger. There is nothing to show whether this was the original arrangement or a restoration.

¹⁶⁹ In the Athenaeum the thickness of the walls, both of the upper and lower enclosures, varies (approximately) from 6 ft. to 7 ft.; with the exception of a small part of the walls of style (3), where it is only about 5 ft.

At Sellasia the original thickness of the walls is, in their present condition, very difficult to determine. There is no doubt that that of the outer wall varied considerably in different places;—in some it was only about 9 ft., while in one place I measured 12 ft.; the average thickness was perhaps about 10 ft. The average thickness of the wall which separated the inner enclosure from the outer was about 8 ft.

The plans given in Figs. 6 and 7 are necessarily incomplete; for, as I have already noted, I failed to find any certain traces of the entrances to either fort; and, in the case of the Athenaeum, it is quite possible, perhaps even probable, that some of the towers have wholly disappeared. In one or two places I have had to indicate the supposed position of the walls by dotted lines.

The contour-lines show with tolerable accuracy the relation of the walls to the natural contours of the ground, the level of each of the towers having been ascertained, previous to their insertion, by trigonometrical observation. Elsewhere the contours are only roughly sketched.

¹⁷⁰ Leake, , Peloponnesiaca, p. 238Google Scholar.

¹⁷¹ My authority for the modern name of this river is the French map. I do not think I have actually heard it used.

In the French map this river is marked as the Brentheates (Paus. v. 7. 1; viii. 28. 7); but this is inconsistent with the generally received and almost inevitable theory which places Brenthe near Karýtaena. Besides, Pausanias (viii. 28. 7) distinctly says that the Brentheates, after a course of only five stades (rather more than half a mile), ran into the Alpheius. It is doubtless, therefore, the short but copious stream which rises just below Karýtaena and turns the mill close by the Frankish bridge (cf. Boblaye, pp. 164, sq.).

¹⁷² These remains are marked in the French map, but (unless there are other remains which I have failed to see) they are marked on the wrong side of the stream. I have accordingly corrected this slight error.

¹⁷³ In the French map a track is marked along this valley leading from Stemnítsa to Tripolitsá.

¹⁷⁴ Peloponnesiaca, pp. 238, sq. The form Zibovísi, which Leake uses, is erroneous.

¹⁷⁵ Leake, Pelop., map; Boblaye, p. 172. On the other hand, Ross, (Reisen im Pelop. p. 120Google Scholar), followed by Curtius (i. 315) and Bursian (ii. 229), places Soumetia at Palaea-Sylímna, which is a totally different place, lying considerably west and somewhat north (instead of south-west) of Sylimna. It is a very steep hill, overlooking the plain of Daviá, and containing on its summit remains of fortification, church, and other walls, but nothing necessarily dating from ancient times. Quite close to the village of Sylímna is yet a third hill, of much smaller dimensions, surmounted by remains of fortification-walls. This is the ‘Palaeó-kastro of Sylímna,’ and contains nothing Hellenic.

¹⁷⁶ The upper part of the Helisson goes by the name of the ‘River of Daviá.’

¹⁷⁷ Ross, p. 118, sq.; Curtius i. 315.

¹⁷⁸ Leake, , Travels in the Morea, ii. 52Google Scholar. Bursian (ii. 228) does not attempt to decide between the claims of Maenalus and Dipaea. For the form ‘Dipaea’ v. Paus. viii. 27. 3.

¹⁷⁹ Herod. ix. 35; Paus. iii. 11. 7; viii. 8. 6; viii. 45. 2.

¹⁸⁰ Paus. viii. 30. 1.

¹⁸¹ Morea ii. 305, sq.; Pelop. p. 241. sqq.

¹⁸² Possibly the marsh (ἕλος), which was formerly here, has been completely absorbed by the ever-widening river-bed. Just east of the town, where the marsh was, it is of great width (v. Excavations at Megalopolis, Pl. I.).

¹⁸³ Pelop. p. 234.

¹⁸⁴ Morea ii. 297, and Pl. 2.

¹⁸⁵ Morea ii. 44; Pelop. p. 235.

¹⁸⁶ Morea, loc. cit.

¹⁸⁷ Leake, , Morea iii. 18Google Scholar, sq.; Bursian ii. 114.

¹⁸⁸ P. 45.

¹⁸⁹ P. 235.

¹⁹⁰ The valley of Akhladókambo lies some 1200 ft. lower than that of Tripolitsá.

¹⁹¹ Bulletin de Corr. Hell. vol. xiv. (1890), p. 382.

¹⁹² Guide Joanne, p. 236.

¹⁹³ Herod. vi. 105, 106; Paus. i. 28. 4, viii. 54. 6.

¹⁹⁴ viii. 54. 7.

¹⁹⁵ P. 35.

¹⁹⁶ Assuming, what is almost certain, that the route described in Paus. ii. 25. 1—3 as ‘the’ route from Argos to Mantineia is identical with the ‘Prinus’ route of Paus. viii. 6. 6 sqq. But in any case it is distinctly stated in viii. 6. 6 that the Prinus road did not follow the Inachus except so far as that river formed the boundary between the Mantineian and Argive territories, i e. up in the hills.

¹⁹⁷ Both identifications are of course conjectural. There is no trace of a κρήνη (supposing κρήνη to imply an artificial basin or fountain of spring-water), as distinguished from a πηγή, at either place. Where I have marked the fountain Arne the spring-water does not come to a head (at present) at any one point, but (as the rustics say) ‘βγάζєι τὸ υέρος,’—the whole place runs with it.

¹⁹⁸ Morea, vol. iii. p. 53.

¹⁹⁹ I have disregarded Leake's, suggestion (Peloponnesiaca, p. 371Google Scholar), that the ‘Prinus’ and ‘Klimax’ routes coincided, on the Argive side, as far as Oenoë, which he places near the modern Kato-Bélesi (Katobélissi), though it receives some countenance from Paus. ii. 25. 1, where only one route from Argos to Mantineia is mentioned. Leake's view would imply that the ‘Prinus’ road followed the Inachus all the way from the Argive plain to near the top of the pass, which we know that it did not; and it makes the ‘Prinus’ an extremely roundabout and unnatural route (v. map of the Mantinice and Tegeatis at end of Peloponnesiaca).

²⁰⁰ Curt. Pelop. vol. i. Pl. III.

²⁰¹ V. Guide Joanne, p. 379.

²⁰² Cf. Curtius, , Peloponnesus, vol. i. p. 246Google Scholar.

²⁰³ viii. 11. 7.

²⁰⁴ P. 372.

^204a Pelopnnesiaca, pp. 230, sq.

²⁰⁵ We have already, in the course of the present paper, seen more than one instance in which that rule was not observed.

²⁰⁶ Now called, τοῦ Τουρνικιώτη, but the old name is still remembered.

²⁰⁷ Vol. xiv. (1890), Pl. I.; reproduced in the Guide Joanne, opposite p. 374.

²⁰⁸ Notwithstanding also that both ‘Ptolis’ and the tomb of Penelope are mentioned on one only of Pausanias' routes; and, as Gourtsoúli probably represents either ‘Ptolis’ or the tomb, it must (supposing the routes to have gone on opposite sides of it) have been equally near to both, But it is quite in accordance with the methods of Pausanias, when describing two routes, to mention the objects on the second only in so far as they differ from those already mentioned on the first.

²⁰⁹ Paus. viii. 12. 7.

²¹⁰ The tradition that it was a tomb perhaps accounts for the expression which is applied to it. ‘Ptolis,’ though smaller, is called an For (Paus. viii. 12. 7) Leake, (Pelop. p. 381Google Scholar, note) suggests Cf. Paus. viii. 44. 7, where a somewhat similar hill (Kresium; v. supra, p. 35, note 36) is described as an

²¹¹ Vol. iii. pp. 57—93.

²¹² Morea, vol. iii. pl. 2.

²¹³ Thuc. v. 65.

²¹⁴ Pp. 380, sq.

²¹⁵ Recherches, p. 140

²¹⁶ The object of Leake and Boblaye was, apparently, to find a larger stream than the one which now flows past the walls of the ancient town.

²¹⁷ South as well as west; for the Orchomenian contingent, in order to join him, had to pass Mantineia (Xen., Hell. vi. 5. 17Google Scholar).

²¹⁸ Morea, vol. iii. p. 75.—It must be remembered that the ‘Argon’ there referred to is the plain of Louká, so that the ‘smaller and more northern branch of the Mantinic plain between Mantineia and the Argon’ probably means our ‘Argon,’ i.e. the plain of Tsipianá.

²¹⁹ Paus. viii. 11. 10.

²²⁰ Id. 11. 7, 8; 12. 1.

²²¹ xv. 85, 86.

²²² Leake, (Morea, vol. iii. p. 81Google Scholar) apparently supposes Epaminondas to have been personally engaged in the cavalry engagement which preceded the battle, and never to have returned to Tegea. This view is countenanced by Diodorus' account of the battle, but opposed to that of Xenophon, whose authority, as a contemporary and a soldier, is rightly accepted as the more reliable.

²²³ Xen., Hell. vii. 5Google Scholar. 21.

²²⁴ Others read and Leake, (Morea, vol. iii. p. 78Google Scholar, note) quotes the passage as The reading I adopt is from the Teubner edition of 1890 (ed. Keller).

²²⁵ Morea, vol. iii. p. 81.

²²⁶ Cf. Grote, , History of Greece, vol. viii. pp. 23Google Scholarsqq. (10 vol. edition) and accompanying plan. My views respecting the site of the battle, formed on the spot, coincide very nearly with those of Grote.

²²⁷ A ‘Graben’ is marked in Curtius' map of the plain (vol. i. pl. III.) and appears again as ‘Fossé’ in the Guide Joanne, being evidently nserted with special reference to this battle. But though the whole plain is intersected with ditches, I can find no trace of this particular one.

²²⁸ P. 379, sq.

²²⁹ Paus. viii. 12. 2.