here and there by a patch of green marsh, where reeds and other waterplants grew. In November the lake began to fill again fast.

Such was the ordinary annual cycle of changes in the Copaic Lake in modern times, and we have no reason to suppose that it was essentially different in antiquity. But at all times the water of the lake has been liable to be raised above or depressed below its customary level by unusually heavy or scanty rainfall in winter or by the accidental clogging or opening of the chasms. As we read in ancient authors of drowned cities on the margin of the lake (Strabo, ix. p. 407; Paus. ix. 24. 2), so a modern traveller tells of villagers forced to flee before the rising flood, and of vineyards and corn-fields seen under water.

Of the places at which the water of the lake, after vanishing into the chasms, comes or used to come to the surface again, three at least can be identified with tolerable certainty. One of them, as we have seen (p. 106), is the ancient Anchoë, the modern Kephalari, in the upper valley of Larymna. The spring which, before the lake was drained, used to issue from the rifted rock was the outflow of the water that fell into the chasms called Bynia (as to which see above, p. 105). Second, the river Melas, which still flows into the Great Katavothra at the northeastern extremity of the Copaic plain, issues again some 4 miles away at the great springs beside the bay of Skroponeri. The way from the katavothra to the springs leads eastward over a ridge of the mountains, and then down a narrow glen to the head of the bay. High mountains clothed with low woods and dense shrubbery, where wild boars used to have their lair, rise on all sides from the shores of this deep, still, fjord-like inlet of the sea; and at their foot, in the inmost corner of the bay, nine copious springs gush from the jagged limestone rocks and pour into the sea, which is so near that its waves wash over the springs and communicate a brackish taste to their water, or rather to all of them except the smallest and most northerly which rises among bushes somewhat farther from the shore. The largest of the springs issues from many rifts in the face of a perpendicular cliff and falls into a basin which used to be enclosed by a wall. The third natural outlet of the Copaic Lake was apparently through a subterranean channel communicating with the lakes Likeri and Paralimni; for the level of the water in these two lakes was observed to vary with the level of the water in the Copaic Lake. Before the Copaic Lake was drained, Lake Likeri stood generally at a height of 45 metres, and Lake Paralimni at a height of 35 metres above the level of the sea, or 52 and 62 metres respectively lower than the winter level of the Copaic Lake. Lastly, it has sometimes been held that the branch of the Melas which falls into a chasm at Stroviki (see above, p. 111) reappears in the copious salt springs called Armyro at Opus. Here, ten minutes to the south of the long south wall of the ancient city, a narrow pass runs between steep rocks and the sea. The springs issue from the rocks in such volume and with such force that they at once turn three mills. In support of the view that these springs are an outflow of the Copaic Lake, it has been observed that at the bottom of an old well in Pavlo, a village to the north of Stroviki, the water is seen to flow from south to north. The geologist Mr. Philippson,

VOL. V

I

however, thinks that the stratification of the rocks through which, on this supposition, the water would have to pass is against the hypothesis.

The plan of draining the Copaic Lake, which has been successfully accomplished within the last few years, was conceived and apparently executed at a very remote time in antiquity. Strabo reports a tradition (ix. p. 415) that the whole basin of the lake had at one time been drained and cultivated by the people of Orchomenus, and this tradition has been strikingly confirmed by the recent discovery of a complete and very ancient system of drainage works in the bed of the lake. The discovery was made by the engineers charged with the execution of the modern drainage works. As described by them, the ancient works were composed of an ingenious combination of dykes and canals, which completely encircled the lake and, receiving the waters of the streams which flowed into it on the west and south, conducted them to the chasms on the east and north-east banks. Where the canal skirted closely the precipitous rocky shore of the lake a single dyke or embankment sufficed, the water being led between the dyke and the shore. But where the canal had to cross a bay, or where the bank of the lake was not high and steep enough to serve as one side of the canal, two parallel dykes were constructed and the water flowed between them. The remains of these ancient drainage works in the bed of the lake are of two sorts. In the first place we see them as low broad mounds, about 5 feet high and 50 to 60 yards wide, stretching for long distances across the plain, either in an unbroken line or with occasional gaps. Sometimes it is a single mound that we see, sometimes two parallel mounds at a short distance from each other. And between the two parallel mounds or beside the single one a long shallow depression marks the line of the ancient canal. These long low broad mounds are clearly the remains of the dykes which formerly enclosed the canals, and which have been gradually reduced to their present level by the ceaseless wash of the waters in the course of ages. In the second place, the line of the ancient canals may be traced by the walls built of great polygonal blocks which in many places support and case the inner side of the dykes. In some places these walls are well preserved, but in others nothing of them remains but a conspicuous line of white stones running for miles through the otherwise stoneless plain.

There were three of these ancient canals. One skirted the left or north bank of the lake, another skirted the right or south bank, and a third ran out into the middle of it.

(1) The first of these, the northern canal, began on the west bank of the lake, at the modern village of Karya, in the shape of a single dyke, which extending north-eastward dammed up the water of the Cephisus and conducted it to join the Melas at a point between Pyrgo and Stroviki. No attempt seems to have been made to drain the marsh in the bay of Tzamali, where the Melas rises. The water of the united rivers was then conducted along the north bank of the lake behind the island of Stroviki and the peninsula of Copae (Topolia), in a canal formed by the cliffs on one side and a strong dyke on the other. From Copae the canal, now enclosed by an artificial dyke on both sides, struck

north-eastward across the great bay of Topolia to the rocky cape at Pyrgos H. Marina, where it was joined by the southern canal. From this point the united waters were led eastward in a single canal for about 2 miles and then dispersed among several branches which conveyed them separately to the various chasms. Of these chasms at the north-eastern corner of the lake the most important are those which go by the name of Spitia (or Varia), Bynia, and the Great Katavothra. The branch canal which led into the Spitia chasm is especially well preserved. It is to be observed that the Melas river, which no doubt originally flowed in the great northern canal from near Orchomenus to the north-eastern corner of the lake, now parts company with the canal at the former island of Stroviki and thenceforward flows to the Great Katavothra in a deep bed of its own.

(2) The central canal received the waters of the Hercyna river, which were led to it by means of two dykes converging on the mouth of the canal like the sticks of a fan. For a length of 4 miles this canal, formed by two embankments without masonry, has been traced running out into the middle of the plain. But the embankments gradually diminish in height till they wholly disappear in an impenetrable jungle of reeds near a large barrow or tumulus surmounted by a block of marble. Beyond this point no vestige of the canal has been found. The French engineers who drained the lake think it likely that the canal was continued right across the plain and joined the southern canal on the eastern side of the lake near Karditsa. But to this it is objected by Mr. Philippson that it would have been a needless expenditure of labour to construct a canal for the waters of the Hercyna across the whole breadth of the plain, and that too the highest part of the plain, when it would have been easy to turn the water of that stream into the southern canal, thus making one canal do instead of two. He therefore plausibly suggests that the water of the central canal, after being conducted to the middle and highest part of the plain, was thence dispersed in a multitude of tiny rills and channels for the purpose of watering the fields.

(3) The southern canal was intended to collect the waters of the streams which flowed into the lake from the south, the chief of these streams being the Phalarus and Lophis. It begins at the mouth of the Phalarus, beside the village of Mamoura, and runs eastward till it is joined by a branch canal from the south. Both these canals, curiously enough, are embanked only on the right side. Soon after their junction the canal disappears in a deep natural depression which extends along the southern side of the plain and comprises the marsh of Haliartus. To the east of Moulki the canal begins again close to the bank, where it probably received the waters of the Lophis (river of Moulki). Mr. Kambanis, the French engineer, thinks that the canal was originally continuous all along the southern side of the plain, and that the missing central portion of it has been destroyed. Mr. Philippson, on the other hand, conjectures that the western portion of the canal ended in the marsh of Haliartus, where the water may have been either suffered to evaporate or dispersed in rills to irrigate the fields. However that may be, the portion of the canal which begins to the east of Moulki skirts the