The problems presented by the trieres are manifold. I shall here be concerned mainly with structural matters. Overall length and breadth are governed by the shipsheds which we know to have been built for triereis. These have less than metres between the columns ( ft ins) and a dry length of metres ( ft ins). Since there were thranite oars a side, a minimum length can be obtained by taking x ft for the rowing space and adding to ft for bow and stern. The length is therefore likely to have been about ft. Breadth amidships is likely to have been ft, ft including outriggers.

The th century representations appear to show a galley with an overall deck under which sat oarsmen at three levels. The naval lists give three classes of oars thranite (), zygian (), and thalamian (), and the literary texts confirm these names for the three classes of oarsmen. One hundred and seventy oarsmen and petty officers and marines seems to have been the standard complement. Representations show that the upper level rowed through an outrigger, while the lower levels rowed through portholes, although a porthole is only necessary for the lowest of all. The relation of the lower levels to each other seems to have been exactly as in the two-level ships, the addition of the outrigger being the specific differentia which constituted the trieres. The overall deck seems to have been an East Greek innovation which was

not adopted by Athenian triereis until after the Persian wars. For the Athenians at Salamis the chief offensive weapon was the ram and their tactical dispositions were governed by the use of it. They took on board only soldiers and archers, while the East Greek ships, which were fighting for the Persians, carried as many as soldiers on their decks. This practice seems to indicate a different tactical theory. Confirmation is to be found in Plutarch's account of Cimon's naval operations in Ionia in BC. He says that Cimon took over a fleet of triereis which `had from the beginning been well constructed by Themistocles for speed and manoeuvrability' ie the Salamis fleet. `He made them broader and put a bridge between their decks so that they might be able to attack the enemy in a more formidable fashion with many soldiers'. We may note that the Chians, who appear to have favoured this theory of naval warfare, were sailing with him on this occasion. Cimon's political tendencies may have inclined him to a tactic which gave greater importance to the soldier than to the oarsman, but for whatever reason he seems clearly to be giving up the Themistoclean tactic which saw the ship primarily as an oar-powered machine for ramming and sinking the enemy, and adopting the opposing theory by which the ship was a platform for the accom­modation of as large a force of soldiers and archers as possible. The phrase 'making the ships broader and putting a bridge between the decks' must indicate the linking of the bow and stem platforms with a continuous deck which overhung the gunwale. The overhanging deck is suggested in the representations. In the realm of naval theory as of politics Cimon represents a backward eddy in a stream which was, for more than a century, to flow the other way. His modification of the Athenian trieres proved to be perman­ent. The Athenians did not however adopt his increase in the number of deck soldiers, but retained the same number which had been used at Salamis. Thucydides refers to the Cimonian tactic as the old­fashioned way of fighting a sea battle.

The general appearance of the Greek and Persian fleets at Artemisium seen from a distance seems to have been very similar. The silhouette from ahead or astem would not differ whether or not the overall deck was present, nor would this feature make much difference in the silhouette from abeam. Some of the Persian fleet mistake a Greek ambushing force for their own ships. Yet there were differences in perform­ance, which are ascribed sometimes to differences in the quality of the crews, sometimes to differences in the ships themselves. At the beginning of the Ionian revolt the Persian fleet in Cyprus was manned by Phoenicians and was defeated by a better-trained Ionian fleet in which the Samians were pre-eminent; and later, after Salamis, the Samians, confident in their superiority over the Persian ships, bring the Greek fleet under Eurybiades to Mycale. On the other hand at Artemisium the Persians are described as being aware that their own ships, besides being many times more numerous, were also `better movers in the water'. In the debate among the Greek captains before Salamis reported by Herodotus Themistocles argues against retreat to the Isthmus..If they fight there `it will be in the open sea, which is least advantageous to us since our ships are heavier and fewer in number'. Among the Greek ships those of the Athenians `moved best in the water'. We can only guess at the reason for a ship being light or heavy, moving well or badly. It seems unlikely that the reason lay in faulty shipbuilding, in different design, or in lack of skill in rowing. The main reason certainly lies in a condition of the ships, which could be, and was, accepted without shame or apology by Themistocles. It seems likely that the clue is contained in Herodotus' account of the approach of the Persian invasion fleet: `the naval commanders, on orders from Xerxes, moved all the ships from Doriscus to the adjoining beach . . . Here the ships were all hauled ashore and dried out'. The 'drying-out' operation no doubt included caulking. With the initiative of the attacker the Persians were able to arrive in Greek waters with dry bilges, while the Greeks, never knowing when the Persians might appear, could not risk taking their ships out of commission for a similar drying-out opera­tion.. In the crucial battles, then, the Greek ships were heavy compared with their adversaries.

Although the Athenian ships at Salamis moved more slowly in the water than those of the Persians and their Greek allies, in other respects they appear to have been more manoeuvrable. Plutarch's account of Themistocles' tactics on this occasion is interesting. Themistocles took care, he says, not to send his triereis bow-on against the enemy until the time arrived when a fresh breeze from the sea and a swell through the straits usually arose. `This breeze had no effect on the Greek ships, which had a low centre of gravity and lay low in the water, but was fatal to the Persian ships which had high stems and decks and great momentum when under way. It caught them and offered them broadside to the enemy'. Plutarch's evidence has been thought to clash with Herodotus' statements about the relative condition of the fleets. We saw that Herodotus is probably talking about the state of their hulls as affecting their speed under oar in open water, the Persians having had the advantage of drying out while the Greeks had not. Plutarch on the other hand is talking about the manoeuvrability of the Persian ships as a result of their build, and about their ability to keep station in a stiff breeze. The presence of the overall deck with or extra men on it would be likely to have just the effect Plutarch describes.

In conclusion it may be of interest to summarise the literary evidence for the structure of the hull. In Aristophanes pitch and timber are spoken of as the main materials of which triereis were made, and pitch is joined with leather fittings and sailcloth in a list of contraband whose export was forbidden from Athens. Theophrastus in the fourth century tells us that ships' keels were made of oak to withstand the wear of hauling up and down beaches. Internal woodwork, he says, was sometimes made of pine for light­ness, otherwise of mulberry, ash, elm or plane. Plato in the Laws gives full information about the rest of the wood used. The Athenian stranger asks: `How is our district off for timber for shipbuilding?' Clinias replies, `There is no fir to speak of nor pine and not much cypress; nor could one find much larch or plane, which shipwrights are always obliged to use for the interior parts of ships'.' The use of pitch is presum­ably for caulking seams and generally making the ship watertight. In a miscellaneous list of naval stores in an inscription of we find hupaloiphe, outside paint, in two varieties, `white', and `black'. The black is presumably pitch. It seems likely that a new coat of pitch was put on before each new sailing. The `white' hupaloiphe may be just clear varnish of some kind, possibly wax with which Hipponax (the century) appears to have said the keel of a ship was smeared, possibly tallow.

Although pitch was used generously on triereis' hulls, they seem to have leaked water into the bilges fairly quickly. We have discussed this question in relation to the respective speeds in the water of the Persian and Greek fleets before Salamis. The Athenian ships at Syracuse became heavy and their speed reduced months after commissioning. The first thing that Lysander did on his appointment to command the Spartan fleet in Asia Minor in BC was to beach the ships and dry them out.. The function of the ship-shed was presumably to keep the hull in good condition out of the sea and sheltered from rain and sun. We learn of some ships which had become leaky through having been out of the sea for too long.

The idea of holding a symposium on the theme 'the history of shipbuilding' could hardly have arisen at a better time than during these years, when maritime archaeology has begun in earnest to contribute to research with primary source material: the wrecks of ships, whose constructional character­istics we had only a very vague conception of earlier. It is important for this material to be examined and described with great care, in order to nourish scientific debate and enable a fresh and deeper understand­ing of the many ship types which have, during the centuries, influenced the course of history.

In contrast to the general character of the title the subject of the present paper is the presentation of a single find: the ships from about the year which blocked a channel in the Roskilde Fjord in Denmark. The Norwegian burial ships from Oseberg and Gokstad have been described so often that a closer presentation of these is presumably superfluous. While other Viking ship finds, though numerous, are so scattered and often in such a poor state of preservation that they do not illustrate the range in the shipbuilding of the Viking Age to quite the same extent as the ships found in the Roskilde Fjord.

The site was in one of the channels in the fjord, the Peberrende, where the ships were blocking the passage to the cathedral-town of Roskilde. The blockage was mapped underwater in - by the Danish National Museum and the draining and complete excavation was carried out in the summer of . In the course of a few months we were able to uncover the remains of the ships from below the stones, raise every fragment of the fragile planks and bring the wood safely to the ship conservation workshop built up for the occasion in an old factory at Brede north of Copenhagen.

To outward appearances, work in the conservation hall was not surrounded by the same sense of drama as that in the fjord, but the task to be accomplished was certainly gruelling enough. For it now remained to be seen whether we could manage the thousands of fragments comprising the wrecks, conserve them effectively, and reassemble them to give an accurate impression of the ships.

It soon became apparent that full scale drawings of all planks etcetera would have to be made, in order to keep a check on the conservation process, and the reconstruction to follow. After being recorded, the fragments of each plank were assembled on a long table; a sheet of glass was placed over the plank, and this is turn was covered with a length of transparent polythene. The outline of the plank with all details could then be traced on the polythene sheeting, and a full scale drawing of it obtained. Afterwards the fragments were packed in glass-fibre gauze, and put aside for conservation. It took years - with only few interruptions - before the task of recording the planks in this manner was completed, on more than kilometres of polythene.

Each piece of timber and every plank was examined closely at Brede, in order to see how the wood had been hewn, and what tools had been used to dress the ship's timbers. We found countless axemarks and plane traces but no sawmarks. It was clear that methods were adopted for cleaving logs intended for planking: in one, the log was split in half, and each was fashioned into a plank. This was the method used for pine planks, and for a few, very broad oak planks. All the remaining oak planks were made by splitting the logs in series of diminishing halves - that is to say '/Z, 'a, ', /, and `/,,. These wedge­shaped elements were then fashioned into planks. We know the tools used for cleaving wood in this way from several finds and medieval representations.