¹ The following works, to which constant reference is made in the text, are quoted subsequently by their authors' names only:—Conze, A., Reise auf den Inseln des thrakischen Heeres, Hannover, 1860Google Scholar; Perrot, G., Mémoire sur ľisle de Thasos (Archives des Missions Scientifiques, Tome 1^er, 2^e série, 1864)Google Scholar; Fredrich, C., Thasos, Athenische Mitteilungen, vol. xxxiii. (1908)Google Scholar.

² The outline of this map has been redrawn from the British Admiralty chart which, excellent as to coastline, has been the cause of much subsequent error owing to the capital of the island, Theológo, being marked in entirely the wrong place. The magnetic variation has been corrected to 1907.

³ Schinas, , Μακϵδονία, Γ. p. 845Google Scholar, gives only two streams as permanent: that of Potamiá in the N.E., and that which flows between Kastro and Theológo to the S.W. To these, however, should be added the streams of Panagiá and Liména.

⁴ This has made me decide not to insert the tracks connecting village with village. These it would be impossible to insert accurately among contours which give only a general idea of the country, and they lie for the most part through thick woods, thus adding to the difficulty of correct observation. Moreover the traveller is invariably dependent on mules who kuow every inch of the road.

⁵ ῾᾿Ηδε ὤστ᾿ ὄνου ῥάχις

῾´Εττηκεν ὕλης ἀγρίας ἐπιστεφής

Quoted by M. Perrot, p. 14.

⁶ I wish gratefully to record obligations to H. E. the Vali Pacha of Salonika and H. E. the Mutessarif of Thasos; to Mr. A. W. Graves and Mr. Mullock of Salonika, and Major A. L. Stephen Bey and other officers of the Gendarmerie Internationale Impériale at Kavalla, for invaluable assistance and generous hospitality; to M. Edwin Speidel for help and advice on topographical points, and to Messrs. G. F. Hill, J. H. Hopkinson, and A. J. B. Wace for their good counsel in writing this article. In speaking of the village of Panagía I express my obligation to many kind Hellenic friends in the island, but I mention here my faithful friend and comrade, Ali Ismail, a Cretan gendarme, who was with me in every possible capacity unrequited for five months and showed an ability and integrity which would do credit to any service.

⁷ This map differs widely from M. Perrot's plan (Pl. II.), which is, to speak frankly, very inaccurate. In its general lines it resembles far more closely those given by Dr. Conze and Herr Fredrich, but contains considerably more detail. The modelled contours were made with a fairly close acquaintance with the ground, but are not scientifically accurate. They have the advantage of giving a good general impression of the aspect of a fortified Greek city facing seawards.

⁸ Skylax, 67: λιμένες δύο τούτων ὁ εῖς κλειστός

⁹ In a country where food can be a real difficulty fish-spearing at night with a fire of resinous pine chips is one of the best means of replenishing. Fishing with nets is, as usual in the Mediterranean, most uncertain.

¹⁰ This should be slightly more to the N. than it is in the plan.

¹¹ Cf. Fredrich, p. 228. I formed the same opinion, though I was told that the pier running due W. was built at the time when the island was attached to the pashalik of Egypt. Should this be so perhaps what I have called a landing-stage was a mole.

¹² Which I learn from other travellers is called ῾Εβραιόκαστρο though I myself never heard it so named.

¹³ It is easy to understand the destruction of the corresponding portion of the W. wall, which runs through the occupied plain, but the wholesale removal of this remote and, judging from masonry near the projecting tower, massive defence requires explanation. It may have been taken to afford material for the adjoining theatre, if this is, as I believe, a late construction.

¹⁴ The first and last word are easily read from the ground with a glass; the curious second name indecipherable from below, I recovered from a squeeze made on an olive ladder.

¹⁵ The traces of 11 are hardly as clear as they appear on the plan.

¹⁶ I have indicated on the plan the ruins of this as they now appear, but it is quite possible that the building was originally rectilinear and that the diverging angles there noticeable were caused by the same upheaval which diverted the course of the ancient bastion below.

¹⁷ Louvre Catalogue, Nos. 704, 705. Dr. Conze took them to belong to the orientalising period of Hellenic Art, while M. Perrot passed them by as the work of a Byzantine artist. A photographic publication dealing with this point would be of interest.

¹⁸ I visited these coves but once and that on the last day of my stay and my note-book may be at fault. In one, I think Pl. XIV. 30, are traces of a rock-cut stairway and quarry. In the other, I think Pl. XIV. 31, a fortification wall. But I may have inadvertently transposed the numbers.

¹⁹ The other occurs just to the E. of the tower at the foot of the steps (see below, p. 218).

²⁰ The traveller should be well booted, as snakes abound. The larger are harmless, but the smaller are too like adders to be good company.

²¹ In Crete water thus obtained is kept sweet by covering the floor of the cistern with white sand: the key of the cistern is the mark of the master of the house. It is considered better to have two cisterns, so that one may be used during the annual changing of the sand, and in effect two are found on the Acropolis of Thasos, 35 and 36 in the plan.

²² This should be entered with the greatest caution as some of the stone beams of the roof are practically now without support.

²³ But the site of the ancient capital is one of the few places in the island where a refuge of this sort cannot for long be maintained, for the Acropolis group is practically an island, and there is but one quite low connecting ridge between the hill marked 25 in Pl. XIV. and the foot hills of the mountain system to the S. of the city. Hence Panagía formed a far safer refuge to the S., and doubtless the mediaeval settlement between the Acropolis and temple was abandoned in favour of that.

^23a It is, of course, true that in the architecture of all nations instances can be quoted of buildings which, by some unrecorded change or chance of history, have never been completed. But the proportion of these in Hellenic architecture cannot fail to strike the traveller.

²⁴ Herr Fredrich in his plan gives 38 × 16 m. as the measurement. My actual measurements were 37·30 × 13·80 m. The measurements are difficult to take owing to the intervening débris and bushes, but if the dowel-holes are correct these measurements fit in with a simple and natural scheme of a temple.

²⁵ I believe this to be an early instance of this particular type of masonry. It consists of large square blocks and narrow vertical and horizontal binders, but the stones are larger and the workmanship better than is usual with walls of this design.

²⁶ I repeat here the bibliography given by Fredrich, Herr, Ath. Mitt. xxxiii. 224Google Scholar. Cyriacus, 117, pp. 54 ff. von Osten, ProkeschDenkwürdigkeiten … aus dem Orient iii. 616Google Scholar, Dissert, d. pont. Accad. rom. di arch. vi. p. 190, Conze, , Reise auf gr. Inseln. p. 10Google Scholar, Pl. VII. 2. To these should be added Fredrich, Herr's own account, Ath. Mitt. xxxiii. 225Google Scholar, and Deonna, M.'s good though small photograph, Rev. Arch. 1909, p. 11Google Scholar. I see when going to press that M. Deonna has had for some time in hand a complete and detailed study of the shrine, which is to appear in the Eph. Arch. I hope this, when it is at last published, will bear out my notes and drawings.

²⁷ A certain discrepancy in travellers' accounts of these sculptures is easily accounted for. When I first found the niche I was convinced that the figure of Pan had been removed, nor do any direct photographs I could take show more than a faint suggestion of the figure. The stone is dark and discoloured with lichens, the relief is low and at no time does the sun fall so as to strike the figure. But a cast of the relief which I took, when strongly lighted from the side, gives a fair illustration (Pl. XX. )

²⁸ For such temporary structures we have in the nature of things little record to help us, and I cannot myself imagine a kind of canopy that would fit these supports. Perhaps M. Deonna may have hit upon a more feasible restoration.

²⁹ This is the highest point in the wall circuit and here we obtain an unrivalled view of the disposition of the ancient and modern town. To the N., over the islet Θασοπόυλα is the long flat marsh of the Thracian coast which, from some atmospheric conditions I could not understand, appears perpetually as a shaft of light against the fawning mass of the mountains beyond. Nearer, at our feet, like a map unrolled, lies Thasos old and new, the moles of the little harbour glittering white upon a placid sea, the houses clustered thick upon the foreshore or scattered more sparsely among the gardens and olive-groves that cover the ancient city. Westward the eye follows the fine indented coastline as far as the northern promontory of the island; to the south, across the close-cropped levels of the grazing grounds, the thickly wooded hills begin to rise, hiding the southward track to Panagía and culminating in the bare peak of H. Elias and the higher wood-crested bluff of ’Psario. Eastwards to the right of the Acropolis hill, and far out across the sea, on a clear day the faint outlines of Samothraki, Lemnos, and Imbros can be discerned like low clouds on the horizon.

^29a Conze, Pl. IV. 13.

^29b Conze, Pl. IV. 15.

³⁰ Dr. Conze's transcription is correct: the camera has made the last few letters appear of a disproportionate size.

^30a Conze, Pl. V.

^30b See above, p. 101, No. 24, for facsimile of the letters and references to the articles by MM. Mendel, Deonna, and Fredrich.

^30c Rev. Arch. xi. p. 27.

³¹ In this connexion I note that M. Deonna writes: ‘J'ai fait donner en ce point quelques coups de pioche pour m'assurer qu'il n'y avait pas là ďinscription comme sur le côté opposé” (op. cit. p. 31).

^31a Fredrich, , Ath Mitt. xxxiii, p. 219Google Scholar.

³² The dotted outlines in the plan are purely conventional.

³³ ὄς κατέκειτο ἐπάνω τοῦ ῾Ηρακλείου Hippocrates, ed. Littré, ii. p. 698. See Conze, p. 16.

³⁴ This is in effect a small Turkish hamlet occupied by the families of the soldiers above.

³⁵ J.H.S. viii. (1887), p. 437.

³⁶ Id. p. 426.

³⁷ In the plan this point is inadvertently indicated by a rectangle, which gives the erroneous idea of an existing foundation. Such may exist but is not visible on the surface.

³⁸ Before leaving the interior of the city it will be convenient to tabulate the roads, as sketched on Pl. XIV.

1. Along the sea-shore from the metôchi at the start of the longer harbour-mole on the E. to the governor's house on the W. the last house on the fore-shore).

2. From the same place turning off by the Genoese tower, keeping the church on the left, to the gate where stands the stele of Zeus and Kike. Pl. XIV. 15. Here issues the W. route to Panagia.

3. Another route connecting the same two points but following a more easterly course at the foot of the hills.

4. From the little locanda near the modern landing-place past the fountain to the ruined gate in the S. wall (Pl. XIV. b). This is the ordinary route to Panagia.

5. From the church to the fountain and thence following the course of the ancient wall to the extreme W. point of the wall circuit where it leaves Liména for Boúrgaro.

6. A straight reach from the soldiers' quarters, Pl. XIV. 23, to the sea-shore.

The wall circuit may be followed without difficulty throughout its course, but except for the region of the plain it is better to be on foot. The quickest ascent to the Acropolis may be made from the region near the longer harbour-mole, up the ridge which separates the depression containing the theatre from the main cavea of Thasos, and so southwards, along but within the wall-line to the Acropolis.

³⁹ Class. Rev. i. pp. 210, 211.

⁴⁰ The spot is said to be called Archoúda, but I place reliance on no name in Thasos given off-hand in answer to query. One's next informant will be sure to give a different one.

⁴¹ I write from the recollection of a detailed plan shown me by M. Speidel, head of the firm carrying on mining operations in the S.E. of the island.

⁴² I found it difficult to enter the spot correctly on the plan, but I believe it to be correct in its bearing on the tower at the foot of the steps.

⁴³ I have endeavoured to indicate on Plate XIV. the character and extent of the modern occupation of Liména. The purpose of the insertion of these details is sufficiently obvious, but it may be well to state succinctly at this point what might be expected from a properly conducted excavation. Broadly speaking there is little hope of finding either a stratified area or the remains of imposing buildings. On the other hand, the works of ancient art that have come from the island are of so fine a quality that a systematic search for these within the enceinte of the ancient capital seems highly desirable. I append particulars of some likely spots.

(a) The theatre seems to have been rifled rather than excavated by Miller and Bent. It would have to be cleared of the extraordinarily dense growth making it at present impassable, a task of some difficulty; and structurally it might prove of little interest; but the region below and to the west of it should certainly bo examined, (b) A strip of waste land in the region E. of the start of the longer harbour-mole, between the ancient town-wall, which now forms the E. side of the metóchi garden (see above p. 204), and the line of small houses ascending the hill, is marked Gräber by Herr Fredrich (Fig. 2) and here ancient remains are very likely to be found, (c) The cavea formed by the Acropolis, temple, and Pan-shrine hills in all probability hides beneath the vegetation and later detritus the architectural remains of the buildings that once crowned these heights, (d) The well-marked hill on which stands the modern gendarmerie has obviously been denuded, but its N. side, including the region of Bent's triumphal arch, might repay investigation, (e) The whole of the district adjoining the ‘star’ formed by the intersection of the three cross-roads should be investigated. This is all waste ground and lies in the heart of the aucient city: at the spot marked 19 in Pl. XIV. metopes and triglyphs still lie upon the ground, and from the circle of the houses S. of this came, if I am correctly informed, the relief of which a detail is given on Pl. XXII. Outside the ancient city-wall but still in its immediate neighbourhood the following districts call for examination:—(f) The region between the Acropolis and the sea to its N.E. (g) The ridge (Pl. XIV. 24) connecting the Acropolis with the commanding height to the S. (25) and (h) the foothill called Archoúda (26) on the other side of the pastures to the E. of the road to Panagía.

⁴⁴ Conze, p. 22.

⁴⁵ This sarcophagus is 2·67 m. long, 1·34 m. broad, and (including the lower chamber, which is not visible in the illustration), 3·55 m. high.

⁴⁶ Possibly it would be more correct to call it a nunnery, as it is occupied by a solitary nun (the only one in the island), who has a great reputation as a doctor.

⁴⁷ P. 27.

⁴⁸ I made my headquarters here for several months. A traveller who voluntarily seeks a land where there is no recognised accommodation for strangers must necessarily look in a kindly spirit on the village life into which he intrudes unasked. This conceded, I have none but most grateful recollections of the personal kindness I met with on all sides. Since at the close of this paper I give an impression wholly unfavourable of the political and commercial outlook in the island, I particularly wish to record here my sense of the personal kindness shown me by the Mouktars and inhabitants of all the villages, and of these I should naturally name my friends at Panagía first. I must particularly mention my respected old friend M. Christides, of whose services Dr. Conze speaks warmly in his memoir. He was so good as to show me the same kind offices exactly half a century later.

I feel at this point compelled to protest against the disservice done to travellers by harsh criticism of the personal characteristics of the inhabitants in the writings of their predecessors. Every traveller in the Mediterranean will sympathise with what M. Perrot complains of, but it is, I submit, impolitic to print such remarks as occur on p. 68 of his monograph. They are, as a matter of fact, remembered to this day, and were quoted to me on the quay at Liména.

⁴⁹ See above, p. 94, No. 8. This I placed in the little church for safe keeping.

⁵⁰ See above, p. 96, for Mr. Toďs transcription, and below, p. 250, for Prof. U. von Wilamowitz-Moellendorff's suggestion.

⁵¹ I once witnessed a little ceremony here of some interest. At nightfall sime little groups of fishermen were seeing to their nets which were spread out on the beach. A woman with a brazier of incense went round to each group as if to cense them, using apparently the same form of words to each. It must, I think, be unusual for women to take any part in the orthodox ritual.

⁵² The monks in this lovely spot are good farmers and kindly hosts. They all speak Greek but the letterpress of the gaudy European prints of their mother monastery which hang on the walls is in Roumanian. The Metóchi is a little republic without superior.

⁵³ Perrot, p. 85.

⁵⁵ I hope that from the above it is clear that both 4 and 5 lie just above the complicated formation of rock and gulley forming the S. end of the level stretch of beach.

⁵⁴ Herod, vi. 47.

⁵⁶ Fredrick, p. 243.

⁵⁷ Archaeologists have looked for ancient metal workings along this coast in vain, but Herr Edwin Speidel, who is conducting mining operations in the S.W. region of the island, informed me that he had discovered ancient workings near Kínara, but that these contained no trace whatever of gold. I am not sure of what precise point Herr Speidel was speaking.

⁵⁸ There is no authority for the name H. Nikólaos in my map.

⁵⁹ The plan on Pl. XVI., which differs in a marked degree from those previously published (Conze, Pl. II., Perrot, Pl. III.), maybe taken as accurate in outline and measurements.

The view as one rounds Cape Stauró for the S. of the island can be extraordinarily impressive. After a morning spent stumbling among the hot marble rocks of the E. coast of the island, where the vivid green of near larch shuts out all air and view, I got a boat at Kínara and for the long afternoon crept down southwards under the shadow of the shore. An hour it may be after sunset I made the rocks off Stauró. Here and suddenly over the vast waste of silver water, I saw for the first time the lone peak of Athos, 60 miles away, silhouetted against a sky of Chrysoprase.

⁶⁰ Perrot, p. 90.

⁶¹ For transcription of its inscription see Conze, p. 31; Perrot, p. 101.

⁶² Perrot, p. 92 and Conze p. 30.

⁶³ J.H.S. Vol. viii. (1887), p. 434.

⁶⁴ P. 90.

⁶⁵ Though I was keeping a lookout for M. Perrot's capitals I saw nothing of them till the last day of my visit, when a boat came in for a cargo of rosin. This was stacked against the side of my house: when it was removed I found the capitals forming the stairway to my quarters, as shown in Pl. XXII. d.

⁶⁶ Deadman's Rock is a good English equivalent.

⁶⁷ Perrot, p. 88.

⁶⁸ The effect of Ντεμὶρ Χαλκᾶς as seen in the morning light nom a boat below is indescribably beautiful, when the glittering scarp is mirrored far into the opalescent depth below. I do not think that anywhere in the world can be water more crystal clear.

⁶⁹ This, it will be noted, is on one-third of the scale of the last map, Pl. XVI., but the insets are on approximately the same scale in both. The coast outline has been enlarged from the Admiralty chart, and the contours from the Austrian staff map; the details and insets I have added.

⁷⁰ Every effort is made by the good priest to connect the water with the church above, and visits like mine from the sea are discouraged for obvious reasons.

⁷¹ Perrot, p. 95.

⁷² Standing on the tall cape which separates Astrís from Potó we look W. across a scene strangely different from those we have encountered in our wandering in uninhabited regions among the fragments of a forgotten past. In the centre of the semicircular reach now before us lies a fleet of boats, a steamer waits off a small pier, an irregular mass of workmen's shanties lines the shore, the hard lines of mineral workings scar the hill above, and a large white gabled building on a height commands the whole. Such is the transformation effected by Herren Speidel, a German firm holding concessions for the mines of valuable zinc ore which have been discovered here and at Sotéro. Their operations, however, are confined to a small area and the mines are so shut off by the surrounding hills that it is difficult to remember elsewhere in the island that there is a part of it where modern industry is thus strenuously pursued. The mines are on the site of ancient workings, but these, I learn, have been destroyed in the new undertaking. Herr Speidel was good enough to give me valued help in the preparation of the general map, and I have the most grateful recollections of the hospitality and goodness of the engineering and medical staff in a most inopportune case of sunstroke.

⁷³ Conze, p. 35

⁷⁴ Perrot, p. 96

⁷⁵ Conze, p. 36.

⁷⁶ Conze, Pl. III. 4.

⁷⁷ Schinas, a Greek geographer, gives, Μακϵδονία, Γ, p. 844, the form Ἁϵσμάτϵς which I incorporated in my map before I had noticed Dr. Conze's elucidation of the form ‘Mats’ which is in general use.

⁷⁸ Conze, p. 40. I found among the ruins one of the inscribed amphora handles (see above, p. 100).

⁷⁹ Nothing could be more effective than the women's dresses of magenta, crocus, and violet flashing in the light against the shadowed grey of the church wall. But they were one and all bought, I fear, in the bazaar at Salonika, and I saw not a stitch of local embroidery.

⁸⁰ I append at this point an attempt at a chronological sequence of these buildings.

I. Good roughly dressed ashlar, with a tendency to polygonal. Of this period Klima (p. 245) is the only example: this may be contemporary with the earliest work on the acropolis at Liména, viz. the N.E. bastion and the lower courses of the S. E. wall.

II. In this group the stones have their faces carefully dressed and the jointing has been accurately executed. They belong in short to the best period of Greek building. Of this group Αβατσινιά (p. 242) is the best. The detail of the masonry of the N. face of the tower (Fig. 23) shows the nicety with which the blocks are adjusted. H. Basileús (p. 245) is now in ruins, as is the round tower of Βαθιά Ποταμιά (p. 241), but an examination of the scattered blocks gives evidence of the same care. The square tower at Aliki (Pl. XVI. 3), though the vertical faces are rougher, belongs to this group, to which lastly the fortress of Καμινορο χἀῖκο (Pl. XVII. 1) may be assigned, on the strength of the well-cut blocks of the long W. retaining wall.

III. We have next to deal with a group which show no less engineering skill and doubtless all needful accuracy for the purpose for which they were intended, but lack the unmistakable refinement of finish of the previous set. The large round tower of Thumonia (Pl. XVI. 2), the lighthouse at Purgos (p. 232), and probably the fortresses at ᾿στὸ ῾Ελληνικό (p. 233) and ᾿στ᾿ Αὐλάκια (p. 231) form this group.

IV. In this set the stones though still large are but roughly trimmed to rectilinear form, their vertical surfaces are undressed, and they are juxtaposed rather than jointed. H. Joánnes (p. 235) and a considerable part of the square tower at ᾿Αμυγλαδία (p. 244) belong to this set. The oblong fort above the beacli at Aliki (Pl. XVI. 5) may also be assigned to this group.

V. We have lastly to do with a set of fortresses in which the same characteristics prevail, save that the blocks are considerably smaller and generally rather better finished. It compri-es the buildingsat Tsouknida (p. 245), ᾿στὸ Λουτρό (Pl. XVI. 1), and probably the ironstone fort at the S. extremity of the island (p. 244).

⁸¹ Tozer, , Aegean Islands, p. 295Google Scholar.

⁸² This set of coins has been presented by the author to the British Museum. [Edd.]