Saverien, Dictionnaire historique, théorique et pratique, 1758.

The shoreline represents a natural boundary between sea and land; the last sight of home; the first sight of a distant land, perhaps of a new discovery. It was a major hazard for the seaman. For a ship of any significant size (such as those needed to master and exploit the lands - or rather oceans - divided by the Treaty of Tordesilhas) the shore was also a most fundamental barrier to navigation. Ships are only built, and to some extent repaired, on dry land. That barrier presented great problems for shipbuilders and seamen alike, that had to be overcome before any navigation was possible. Cradle (berço, ber) is a good term for the structures created to tend the infant ship on that first short and perilous voyage.

This paper will explore some of the evidence for methods of launching large ships in the pre-industrial world before the age of iron; and also for the even more difficult process of hauling large ships ashore for repair; all at the limit of sea and land, but also of technology. The records are sparse, even contradictory. Knowledge of many critical practical details has vanished, and has not yet emerged in archaeological contexts. It is however certain that the methods of launching large ships underwent a profound transition at the end of the seventeenth century, when cradles and slipways developed to allow the largest ships to slide freely to the water. Before that development, and in many places for long after, ships were laboriously dragged afloat in immense temporary structures, "worlds of timber", the cradles of the title. Considerable resources were required to launch a large ship: men, materials, and probably more powerful equipment than that required for any other contemporary application, varying in extent and detail with local conditions such as tides. The process could take many days to complete. Only after the shipbuilders' triumph over brute forces at "O Limite do Mar e da Terra" could navigation begin.

The paper is illustrated with drawings which should be regarded as simplified representations. Translations given are generally by this writer. Terminology is a problem, but as far as practicable later English terms are used for consistency. Brad Loewen in particular has kindly assisted this study with copies of certain French and Basque material.

Chronologically the present evidence starts in the North, and while the methods differ from those recorded from southern Europe, they are more primitive, and provide points of interest. Two distinct sets of early records can be adduced for launching in northern Europe: late thirteenth century records for English galleys, and for the launch of a small vessel in Flanders in the fifteenth century; and seventeenth century Dutch texts, which contribute further insights [1].

Twenty galleys were ordered to be built around the coasts of England in 1295, and a number of summary accounts in Latin survive from their construction [2]. Typically the building site was set up specially, and perhaps surrounded with a fence for security. Purchases of scaffolding and alder spars for shores are recorded, and launching seems to have been carried out on rollers running on planks down a slipway dug for the purpose - a delf. At one site (Newcastle) eight labourers were employed for four days in wetting cables. One published commentary [3] suggests that this was to shorten the cable to start the vessel moving down the slipway, but if rollers were in use there is no obvious need for very large starting forces: wetting a cable will only move the hull a very short distance and cannot readily be repeated. The description is however reminiscent of later texts describing wetting cables to help lift large vessels off their keel blocks prior to launching.

The launch in Flanders in 1438-9 was of a carvel, one of a pair for the Duke of Burgundy, which were built by Portuguese shipwrights sent for the purpose [4]. The relevent items are: "an estrain that was put under the said carvel when it was lowered to the ground; tallow (sieu) to pay the said carvel beneath, and the planks (ays) on which it was launched into the water; two men who worked for eight days to make the ditch (fosse) where the said carvel was launched into the water; four mariners who hauled upon the ropes to put the said carvel into the water and for the loss of one of the said ropes which was broken". We can see similarities with the accounts above. In some way the hull was lowered from its keel blocks, and hauled down the slipway dug through the river bank after building the ship. Here the hull was lowered onto greased planks, which are a possible link to the later methods of the Netherlands, and argue for some continuity between 1438-9 and the first available Dutch records, from the second half of the seventeenth century.

Van Ijk described in 1691 [5] the new procedures of the southern area centred on Rotterdam, in which the shipyard consisted of a floor of planks something over 3 metres wide and some 40 metres in length specifically to spread the load from the keel blocks. These were set up to a height of about one metre. This height was necessary because some of the frames were set up in advance of planking. The bottom planking had thus to be worked on from below. Witsen described the original method still centred on Amsterdam in 1671 [6], and although he has less to say about the structure of the slipway and stocks it is clear that the keel blocks were much lower - perhaps only 0.3 metres above the floor of the yard. The whole assembly, which had the stiffness and strength of a shell at an early stage, was simply tilted to each side to provide better access to the underside of the planking.

The final development of Dutch methods of launching for large ships, at least for the northern area, is widely recorded in engravings and models from the eighteenth century. Chapman also records the details (for a relatively small vessel) for 1768 [7], where it contrasts sharply with the French and English methods: there is no cradle. The bilge is supported directly on two inclined planks erected under the bilges: inclined both at the angle deemed necessary for the hull to slide, which might be steeper than the original line of keel blocks [8], but also transversely to match the angle of the bilge, giving a dihedral effect. This would provide the primary means of securing the stability of the hull during launching, provided the sliding planks did not move under the loads applied, either vertically, or sideways under the inevitable wedging action. These ways have no need to extend inland beyond the point of maximum section of the hull. Typically this was forward of midships, and the Dutch continued to launch bow-first, so the launching ways were relatively short. The ways were built upon piles of planks set at close intervals, and they were heavily shored laterally to posts driven along the slipway. Interestingly these ways are markedly steeper beyond the end of the building area, but no indication is given of their extension beyond the water's edge. It is almost as though the whole ship is intended to pitch forward into deeper water as the bilge moves forward (indeed this appears to be happening in the frontispiece of Van Ijk's book, the bow of a small vessel plunging, and the stern rearing up from the ways), but at great risk that the stern would then ground heavily dynamically, as buoyancy lifted the bow. It is difficult to see how it could work unless the bilge is firmly supported at least until the stern is clear of the end of the building slip, making the method suitable only for vessels with long flat floors. It also required deep water adjacent the slip; such a site would not serve for hauling ships ashore. Van Ijk himself remarked (the translation is by courtesy of Albert Hoving): "... some years ago it happened that a ship fell over on its side while being launched, the sliding plank being too low and loose. One of the planks was pushed aside, and the ship stopped on the slipway, hanging at the stern, where it burst open almost beyond repair. For this reason (I am told) the Portuguese build their ships with the stern low, to go into the water first" [9]. We will return to this final point later. It is a conspicuous consequence of this method that the hull is effectively supported at only two small areas at the bilges during the launch, requiring great strength of construction, and perhaps unsuitable for large vessels.

Before turning to Iberian methods, a few other primitive methods may be noted. One text describes the preparations for the launch of a "galleon" of some 500 tons built in Poland by Venetian shipwrights in 1571 [10]. It was launched with a cradle made from three specially purchased large tree-trunks (needing six horses to haul each one), and ten smaller "tree-trunks" which were only on loan - and were perhaps rollers, if they were not to be altered. It may also be noted that shipbuilders (like seamen) were itinerant, and must have taken their own local methods with them, but equally may have adopted the methods of others as they observed something different or better. In the case of Portugal, many aspects of shipbuilding can be traced to early employment of Genoese shipbuilders from the twelfth century [11], and this must extend to methods of launching, as we shall indeed see below.

Two parallels may be mentioned from other areas. One recent Greek method for small vessels was to place a pair of tree-trunks under the hull, one under each bilge, and large enough to extend below the level of the keel, and running on rollers. Another method of handling the great weight of the hull during operations such as lowering it, avoiding damage from levering on small areas, and reducing the risk of supports slipping, was to pack sand-bags under it, cut away the original supports, and then burst the sand-bags [12]. As described, this was to lower the hull sideways to rest on a standing way laid at one side of the keel, and reminiscent of the account of 1439 from Flanders. A method observed in Madras about 1850 used coils of rope packed with sand, which were slowly unwound to lower the hull [13]. Such simple practical devices may have been widely used, unrecorded, within more complex operations on large ships - and most notably for the Flanders carvel which was explicitly lowered by otherwise unknown means.

One obsolescent method noted by Ollivier in 1736 [14], was to actually launch the vessel "on its keel". This required the supporting grid to be built up to the keel, and a form of bilgeway was actually fastened directly to the hull on each side, so that the hull was supported along three lines. This has some similarities with the Dutch method. These coites were only removed on first careening of the vessel. Ollivier clearly dislikes this method, which he states was prone to premature movement, and to overturning of the ship. It was also, he adds, damaging to the hull, for lack of adequate support, and much more difficult to restart the launch if movement stopped.

The most useful notice available from primitive Mediterranean methods for launching is Crescentio's, written by 1601. The same methods had clearly spread to Portugal, as revealed by the use of the Italian term vaso in Portugal no later than the first half of the fifteenth century, noted by Carbonell Pico [15], and presumably derived from the older Mediterranean galley tradition: "...so many people joined in the work of putting the galleys on their vasos and launching them that most were launched by hand, without capstans" - a significant phrase, as we shall see. Crescentio had heard a garbled account of the tides of the Gulf of Camboia, which he describes as a great convenience for launching in comparison with the difficulties of the Mediterranean. He then describes the Italian techniques used for galleys and larger vessels, based on the use of articulated bilgeways (vasi), supported from the hull by ropes, and moved on rollers. While it is not a clear description (despite reference to hollow boxes, the vasi are drawn as simple planks on edge, for example), the elements for other cradles are identifiable (Fig.1) [16]. Larger vessels required the intervention of capstans in addition to the heaviest gear normally carried on galleys for their mast and yard tackle, the prodani. The cradle is structurally incomplete: something must have prevented the vasi separating under the hull, presumably ropes.

The next major source chronologically (1616) is Fernandes (Fig.2) [17], who is of course describing the largest cradles of the whole era, for an India nau, an order of magnitude larger than Crescentio's vessels. He calls the cradle e(m)nvazadura, clearly related to the original vasi. Fernandes was evidently a master shipwright, and approached his text from the point of view of a carpenter: the term great might well serve to describe a group of cradles. His drawing, although carefully to scale and containing two projections, is incomplete, and the vocabulary and syntax partly a mystery, but it is clear that he combines his vasos - in this case three lines of them under each bilge - with the piled baulks supported upon them to form cribs (casas - crib being a convenient rendering rather than an exact term). These are interlocked with multiple rows of dragas, which function as daggers, to hold the blocks carrying the hull's weight in their place. Pairs of daggers would in later methods be bolted either side of the poppets fore and aft, clamping them rigidly in line, and it may be supposed that Fernandes' dragas are their forerunners. The origin of the English term dagger for this context was apparently unknown in the early nineteenth century, and it may conceivably be a phonetic corruption from the dragas of these cradles.

In each case the problem is to support the weight of the hull upon the vasos, which form an articulated bilgeway, although the contact at the upper end of the support is steeply inclined at the ends of the hull, and will tend to be pushed sideways. Fernandes notes a difference in height of five palmos de goa (1.23 m) for his cribs across their width, matching the slope of the hull, which would be extremely difficult to fill with stable wedges: his text and drawing are incomplete and details are unclear. One major puzzle is that the layout of the vasos in plan follows the curve of the bilge: so do the cribs; and they appear to overhang the vasos by a considerable margin. By comparison with the later straight bilgeways and poppets and stoppings-up set upon them, this must have made it very difficult to fit the dragas. It does allow the height of the cribs to be minimised while providing maximum lateral and structural support at the bilge. The spread of the bilgeways appears over time to have become progressively less as a proportion of the ship's breadth, which helps to reduce the height of the cribs, stoppings-up and poppets, and the slope of the hull where they meet; but this also changed the nature of the support to the hull, and there may have been a relationship with developing systems of framing, for example, to permit it to happen. Such factors are far more significant in large vessels, but there is as yet no documentary or archaeological evidence to explore this further.

The tendency for the vasos and everything above them to separate is controlled by a heavy cable stretched across under the hull at every joint in the vasos. All the longitudinal strains of dragging the weight of the hull are carried through the vasos by heavy pins, one palmo square linking all the vasos at every overlap. It is possible that this articulation is simply for ease of assembly of necessarily short components, but also that it was necessary to compensate for imperfect groundways. Interestingly, the vasos at the head of the cradle are turned up in an arc, as though to prevent their digging into the ground. Did the groundways not extend far enough, then ? That is certainly the implication of their stated length. It seems from this aspect that the hull was built with the bow towards the water, though it is not explicit (not least plan and elevation differ). Items that are unclear are how the heads of the cribs are restrained, and how all the drag ropes are attached to the hull, and to the machinery - anchors and capstans - necessary to haul the hull and cradle. It is virtually certain that they are comparable to those described throughout the next two centuries in other sources.

Gaztañeta's manuscript of about 1688 (Fig.3) [18] contains a similarly confused and unfinished account for a large ship launched at Colindres in Cantabria. He too was a master shipwright. While the vocabulary is equally resistant to formal translation, it is clear that this method reflects an advance during the intervening 70 years. The cribs of Fernandes' cradle are now substantially replaced towards the end of the hull by individual inclined shores, probably in several rows across the width of the bilge, and all held in place by runs of daggers, bound across the shores by lashings at each intersection. (The upper dagger is drawn as single and is behind the poppets, unlike the later dagger plank. This may relate to the fact that in these earlier methods the poppets were actually restrained by the gammonings below the keel; though in this case these are not actually drawn). The bilge is supported through the central section of the hull on a solid mass of chocks and wedges, which also raised the hull clear of the keel blocks in the final preparations for launching. (What would later be called stoppings-up, longitudinal timbers directly under the bilge on which the chocks and wedges act, are apparently still absent - certainly from the drawing). The bilgeways (bassos, retaining the name though apparently not the form of the articulated cradles) are drawn as single timbers. In practice these and all later bilgeways for large vessels would be made up from numerous lengths of very heavy timbers, all carefully jointed, and made smooth on the underside to ensure that they ran easily over the groundways. In this case the bilgeways seem to be joined by heavy timbers beneath the keel; though heavy ropes are in evidence in the text. There are already dog-shores acting between the groundways and the bilgeways to prevent the cradle moving prematurely; and some of the terminology indicates that driving shores, wedges and levers (palanculas) are all set up to start the cradle moving. These devices are an express statement that the first movement was vital: once moving the force needed was reduced. The building slip appears to be relatively level, and then cambers steeply to the water. Considerable force would thus be necessary to move the cradle initially, and the effect of such uneven support to the hull cannot have been good, leading to severe hogging of the hull during launching (though there seems to be a reference to broad wedges required to support the stern "when the bow lowers"). This may be one reason why ships were almost immediately careened (or in England docked) after launching. There is no evidence here of launching ways extended beyond the immediate building area.

An English account of 1636 refers to the cradles used to launch ships by the Portuguese, actually in Goa [19]. The form of the cradle is not specified, but may be as that of Fernandes: "...new galleon...was launched in a device wherein she was built, called a cradle, which is a world of timber made up and fastened on either side to keep her upright, and so with cables, capstans and a multitude of people, they forced her into the water, the way[s] being first very well timbered and tallowed. There was another on the stocks. they are very long a-doing and issue at excessive rates [cost]". Most large English ships were built in dry-dock, and needed no such vast cradle, whence the amazement at this sight. (Another English account of launching a Portuguese ship is Barlow's, given below). On the other hand, cradles were known in England, even when launching from dry-dock. Butler has the following definition [20]: "a framed piece of timber....brought up and raised all along the outside of a ship by the bilge when she is in dry-dock; and it serves to launch a ship with the more security out of this dry-dock. And in some parts these cradles are also used for the same cause, when any of their great ships are brought only to be trimmed; and they are trimmed in these cradles". This latter remark concerns the grounding of vessels for graving, re-caulking, etc. Smith adds that it was a frame of timber much used in Turkey, Spain and Italy for more ease and safety in launching [21].

The next record to note represents a transition not so much of cradle construction (it is one or two decades earlier than Gaztañeta's), but of recording. The Album de Colbert, anonymous and only datable to the period just before 1677, and reflecting the methods of Toulon, contains a drawing of a launching operation which places the emphasis on the tackle required to haul the ship to the water. It is a perspective view, and finely detailed by comparison with earlier records. It is the first to record the tackle (Fig.4a), and the means of anchoring the hauling forces offshore, capstans on grounded "pontoons" in this case. The heavy grillages of groundways are clearly drawn as forming a plane surface and to spread loads over the rough ground of the shipyard. The bilgeways are again drawn as single timbers. The cradle (Fig.4b) has poppets fore and aft, but they are also continued along the bilge, interspaced between every second crib. The cribs now show more clearly as alternate layers of baulks and wedges, built up to support longitudinal timbers called coutelas, literally cutlasses. One is under the bilge, effectively similar to the later stopping up; and the other is under the bottom, extending the whole length of the poppets fore and aft (and although it is not so drawn, almost certainly having the function of the English dagger-plank, nailed to the hull planking to act as a shole or sole-plate for the thrust of the poppets as the text implies). The function of the poppets is not just to shore up the hull, but to anchor the loops of the gammoning (three to each) which pass under the keel.

The main part of the hauling tackle is a set of very large multiple pulley blocks, fastened either to posts driven at the water's edge, or suspended from the transom on either side of the sternpost (it is a bow-first launch). A second set is similarly set up, but is lighter, and hauled directly by men, not a capstan: its function appears to be partly that of steering the cradle if it deviates, by pulling on one side, and slightly out of the line of launch. No guide ribbands have appeared yet. The text is not descriptive, but a catalogue referenced to numbered items on the drawing [22]. It is conspicuous that the tackle is set up not just to start the movement of the ship, but to act upon it over at least the full length of the ship: it was clearly expected that it would have to be dragged afloat. This emphasis on the tackle is a feature of subsequent French accounts.

The final group of sources to be described are eighteenth century, and may best be classified as a group as by encyclopaedists: they are the product of educated men; published, or intended for publication. These sources enable us to place the key date of transition in French methods between about 1677 and 1692. The cradles themselves have become less massive, more carpentry. The first such item is actually Aubin's Dictionnaire de Marine (Amsterdam 1702), heavily based on Dutch sources.

Ollivier represents the full transition, with his treatise dated 1736 noted above, that provides some of the best practical material of all. He described bow- or stern-first launching as optional, and expects many but not all ships to slide freely, after initial resistance is overcome. He needs drag ropes to stop the ship drifting too far, for example. He does however say that vessels had not always been launched in such a simple manner with sliding cradles. The old method of capstans and tackle (caliornes) had been in use only a few years before, and was indeed still used by some. When ships had started to slide with this old method, the capstans could not keep up; in fact there were all sorts of hazards described for either method.

Ozanne produced a series of sketches about 1765-70, another album in fact, with labelled features but without commentary [23]. They illustrate the process of launching from the construction of the bilgeways on, and including the means to start a ship that would not slide freely. That is, Ozanne represents the methods of Ollivier. The great size of the pulley blocks is clear; they are supported on temporary chocks on the sternpost (no rudder can be fitted in such methods of course). A pontoon fitted with a treadwheel hauls from offshore, and the wedges of the arc-boutant - a raking shore acting on the sternpost - are driven in, to lift and drive the sternpost. Ozanne's is also the first illustration to make it plain that the new bilgeway for a large ship was not actually a single timber, but composed of many lengths, and at least two layers, all coaked and lapped to act as a whole. The stoppings up (ventrières) appear in the central section, above chocks and wedges. The poppets are now vertical (colombiers), but retain their gammonings (liures).

Chapman reproduced a formal drawing of similar arrangements (Fig.5), with some descriptive text, in 1768. It actually represents the launch of the Royal Louis at Toulon in 1692. This launch was stern-first, unlike Colbert's. This is a full engineering drawing, albeit to a small scale. It differs from that of the Album de Colbert: the cribs have been replaced with stoppings-up in their final form, and there is far more precise detail. The method is similar, though: a series of rope gammonings under the keel are used in conjunction with wedges within the stoppings-up to lift the hull clear of the keel blocks just prior to launch.

The most complete text for this drawing however is that of V** in Clairbois' Encyclopédie Méthodique, Marine of 1783. The procedures are more fully set out, including for example the need to allow the ship to adjust slowly to the strains of transferring it from keel blocks to cradle; of wetting the gammoning to increase its tension; drag-ropes to halt the ship when it floated; buoying the cradle for recovery, etc. His bilgeways were separated by struts, and tied together by additional gammoning between them.

The comparable English source is probably Steel's Elements and Practice of Naval Architecture, first published in 1805, from which the English vocabulary has been taken (Fig.6) [24]. In each language the vocabulary of launching is a considerable problem. Bilgeways and sliding ways, standing ways and groundways are confused in different descriptions, contemporary and later; daggers and dagger planks; the parts of stoppings-up, and wedges or slices, all cause confusion, partly because they are transferred from one method to another where the function and arrangement may not actually be exactly the same. Translation tends to confuse matters even further. Early Italian, Iberian and French terms have largely disappeared from any similar context; there simply are no known equivalent terms in English.

Steel calls the launching cradle a "grand piece of mechanism and requires every consideration". One pertinent comment is that cradles of his generation were greatly simplified, with the benefit of experience. He might be summing up the two centuries of development from the "worlds of timber" to the simple, if still massive, carpentry cradle of about 1800.

One interesting feature that only Steel refers to is that the cradle was first assembled piecemeal, cutting each piece to fit, and then dismantled in order to grease the bilgeways and sliding ways just before launching. The grease was a mixture of tallow, oil and soft-soap. This probably gives a truer reflection of the time and effort required to construct these huge cradles, but also of the critical effect of the loss of grease squeezed out during the first slow movement of the cradle. Ollivier refers only to tilting the bilgeways with jacks to grease them. Steel also gives more (and quantitative) details of the bilgeways for a 74-gun ship (using decayed spars in part in his example), and joints between the parts, all snaped to prevent their fouling the joints of the sliding planks, and with all nails in these sliding parts punched a whole inch (25mm) below the surface of the timber - itself a good indication of the violent nature of the movement of the cradle. (Several accounts speak of the risk of fire, caused by the friction). The sliding planks are to be set on a plane sloping 1:96 to 1:48 steeper than the keel blocks (which will be on a plane or cambered slightly upwards). Even in this almost final development of the cradle, the two bilgeways are held together under the keel by "several turns of lashings" between ring-bolts, which are bolted through the bilgeways with fore-locks that can be withdrawn from the ship after launching, so that the two parts separate. Similarly, the inshore keel blocks are left to last after the slices (wedges) are driven, and screws are used to drive the ship if it sticks on the ways, replacing the earlier palans and arc-boutants.

One significant item can be gleaned from these sources. No French or Iberian source as late as 1783 is known to refer to the use of longitudinal sliding planks between the transverse groundways and the longitudinal bilgeways, while English methods recorded from 1768 (Chapman) do specify these features. (In a sense they are the central feature of older methods, and of the Dutch method, using no cradle). This may be a critical factor in reducing resistance to sliding, as it reduces the bearing pressure at every point of the bilgeway by a factor of about two. If the bilgeway was crushed locally by uneven loads, that would hinder its movement, no matter how perfect its original surface. There is no longer an arris at each groundway to scrape the vital grease off the bilgeway as it passes (Fig.7).

Evidence for difficulties in launching, confirmation that large ships did not generally slide freely into the water until the eighteenth century, can be gleaned from a number of disparate sources, where such events evidently cause little surprise. Often, incidentally, these accounts also illustrate the point that foreigners are either writing the account of events, or are involved in the work: it is no wonder that there is a tendency for the methods of launching large ships in Europe to converge during the eighteenth century, reinforcing the gist of earlier instances noted, that must have been commonplace.

A passage of some significance for Portuguese shipbuilding history is to be found in the diary of another English seamen, Barlow. He found himself with others of the crew of the ship Queen Catharine in Rio in 1663 at the launch of the Padre Eterno, one of the largest ships built in the seventeenth century [25]. She was a five-decker, of 43.5 metres keel, and apparently around 2,800 tons deadweight capacity [26]: "...the Governor desired of our commander for to help them with our men and what else we could do for the launching of her [perhaps a reflection of the sheer quantity of heavy tackle necessary], which was to be done on their Christmas Day...But that day she could not be launched, nor in seven more, but on our [Julian] Christmas Eve, betimes in the morning, we launched her off into the water, she being a very large and good ship". This is slightly ambiguous (and says nothing of the detail of the cradle), but it appears that they strove for eight days in all to get her afloat. The other factor is that she only went afloat on the top of a tide - to explain the odd early hour, and it is possible that efforts were simply suspended to wait for a better tide. However, no practical man would try to launch such a vast ship with less than the best tides available, and they expected to launch it on the first day. A week later the tides would be declining. The tide-tables could of course be re-constructed to explore the matter. The account has all the hallmarks of dragging a ship to the end of the slipway, to let the tide float her off the cradle. It just took rather a long time to move her far enough down the sloping slipway for the tide to lift her. Interestingly, Barlow's sketch of the ship at launching shows it facing bow to land, albeit already afloat when drawn, and with a huge flag flying.

Launching still did not always go smoothly even in the Ribeira das Naus, much later than this. There is a record from 1711 of the launch of a large ship of about 70 guns, which took four days to drag into the water, breaking quantities of tackle and hawsers in the process. This was in charge of a French shipbuilder, Chabert, and at the same time English and Dutch were also building in the yard [27].

Does this contradict a different account cited for 1721 [28], or indicate the date of a change of method in the Ribeira ? The Gazeta de Lisboa for 21 November 1721 describes the launch of two 50-gun ships, which proceeded with "the greatest velocity", the Royal family being present, seated in a richly furnished Royal Box constructed for the occasion. The launch was followed by a customary celebration with sweets and drinks. The same commentary indicates that ships were launched with great pomp, and that "no power gave impulse to the fall of the great machine and its cradle....". There are reasons to question the date at which this became viable in principle, but had launching become a scheduled event for Royalty to attend, and can the reported velocity be taken literally ? Taken at face value these two accounts alone would place the date of transition in Portugal between 1711 and 1721, though other conflicting evidence may be noted. There is silent testimony to similar ceremony and watching multitudes in the painting by Noël for a launch in Lisbon in 1789. We might note that this ship was launched stern-first [29].

Duro [30] refers to a manuscript work in the Biblioteca Nacional in Madrid entitled Arte de botar al agua los navios, ascribed to the second half of the eighteenth century, but he does not transcribe any part of it.

We have noted that the heavy pulley blocks near the sternpost are all supported from above, with ropes to the stern gun-ports in for example Colbert. This may reveal a possible key reason for all these early ships to be launched bow-first. With a straight sternpost and a long slender run, the hull is structurally better fitted to support the massive forces at the sternpost than at the bow, which is heavily sloping in several directions. Hawsers attached on the stem would have tended to slip down the stem towards the keel, and the hood-ends of the planking and their adjacent frames might have been disturbed by the loads. We may, then, find that large ships came to be launched stern first only when the cradles and launching methods had progressed so far that the ships usually slid down the ways relatively freely. While they had to be dragged, bow-first was a better method. The change as we have seen occurred in France at least between about 1677 and 1692. A comment on the practical shipbuilding problems associated with dragging large ships is contained in a letter from Corte Real to the Spanish King in 1623, on the merits of three and four-deck ships [31]. Reinforcement of the sternpost with a counter-post is spoken of in the context of needing to drag these huge ships.

We may however note that if a ship was subsequently hauled ashore it was most practicable to do so bow-first, so that the keel remained more nearly parallel to the beach and ways, and reduced the stresses upon the hull, and the length of the slipway (Fig.8). Ropes attached to the hull to carry very large drag forces could then be placed around the sternpost in the same way as for launching. However, the ship had then to be re-floated stern-first. This dilemma and the need to develop techniques to overcome it may have led to a realisation that launching could actually be performed stern-first too. This has dynamic benefits during launching, as the stern has less buoyancy than the bow, and will not lift so violently, thus reducing the load on the end of the keel, and the stress on both ship and slip.

It will be difficult to distinguish in a view of a shipyard whether it was a new ship being launched stern-first, or a ship grounded for repair bow-first, and this may account for some references to Portuguese launching stern-first surprisingly early, in the sixteenth century [32], apparently based on iconographic evidence [33]. One supporting item comes from Van Ijk, who as we have seen wrote in 1691 as a matter of surprise that the Portuguese then launched stern-first. (If we consider Chapman's evidence, it is probable that the French also did so by 1692, a curious commentary upon Saverien's 1758 statement in the opening quotation). Barlow appears to draw the same for 1663, though the issue is far from clear, and merits further investigation.

Bowrey's drawing implying side-launching about 1680 will be discussed below: that is the first and only early case discovered for this study, though it had great merits and would become very common in the nineteenth century.

A ship on its cradle reaches a point on the slipway as it is moved towards the water, however slowly, where the buoyancy of the seaward end starts to lift the hull off the slipway, rotating it about the landward end of the bilgeways. As soon as this happens, the end of the bilgeway is the only point where support is transferred to the cradle and thence the hull. At the same time, most of the buoyancy is concentrated at the other end of the hull. The relatively even spread of loads on the hull and slipway has transformed into severe localised loads, and causes the hull to sag - there is little support from the cradle in this sense. The ship has still to be moved some distance down the slipway before it is properly afloat. Launching a ship stern first will generally assist, as the stern usually draws more water and has less buoyancy than the bow, and the ship will tend to be further down the slipway before it rotates. Correspondingly, this requires a longer slipway. The greater the weight and draught of the vessel on launching, the longer the slipway needed to be, and the further below the high water mark. It was therefore advantageous to launch ships part-built. The ship was lighter to handle, needed less depth of water to float it, less (or no) ballast to make it stable, and was less likely to be strained by its own weight as it started to float and rotate.

Small vessels can be hauled ashore manually when necessary, even when there is no tide to assist, or if it is necessary to carry out more protracted repairs than can be achieved between successive tides. Ships the size of caravelas, or rather larger vessels if they are designed to be so treated, and are being grounded on smooth sand or mud, with no risk of settling on rocks, or falling over, may be grounded in tidal areas relatively easily.

Classical galleys, which are estimated to have weighed around 30 tonnes, similar to a small caravela dos descobrimentos, were so hauled ashore in the almost tideless Mediterranean. Theophrastus' Enquiry into plants, V, vii, indicates that the keels of triremes were made of oak, so that they could withstand the abrasion from being hauled ashore regularly. Merchant-men had keels of fir, and if they needed to be hauled out had an oak plank placed to protect them.

Some of the issues associated with grounding ships were considered in an earlier paper [34], and will not be repeated here, but the point is made almost perfectly by a passage from Zurara's Crónica de Guiné [35]. Antão Gonçalves' caravela left the Rio de Ouro about 1441, and immediately it was "seen how his caravela needed to be repaired, he had it put ashore, where he made it clean and repaired what was necessary, waiting on his tide, as was done before the port of Lisbon, at which daring many were astonished".

Alexandre da Fonseca states that ships to be repaired in Lisbon were put aground in cavas along the shore, though no other reference to this has been found. Presumably these cavas were similar to the various forms of early "dock", more or less permanent and either with or without gates, that occurred in England from the fourteenth century [36]. Trueba cites a Spanish text that indicate that it was quite normal to haul ships ashore and even to raise them onto stocks for repairs to the keel about 1535: "sean barados en tierra e puestos sobre picaderos de manera que descubran toda la quilla" [37]. As India naus grew in size in later years, only the greatest tides would serve much purpose in simply grounding such ships before the port of Lisbon.

Hauling ashore is not just the reverse of launching: there are a few additional difficulties to note. Ships to be hauled ashore were complete, and probably waterlogged from long service, and were thus much heavier than when they were launched. Part of the force to be overcome was that of gravity, which had assisted in launching, though it is an advantage that the force is directly ashore, and does not have to be applied towards the water.

While a launching cradle can be carefully constructed to fit a ship for launching, and fell away freely as the hull floated, there is much greater difficulty in placing a cradle under a ship afloat, that will properly match its form and provide full support. The old style of rope-based cradle might be doubly difficult, as there was no rigidity between the two sides until the ropes were tensioned against the weight of the hull. Vessels of any size will distort with service, usually hogging, as their joints are worked by the alternating loads on them. That is, after a period at sea the ship becomes deformed with the keel deflecting upwards into an arc, perhaps half a metre. A decision has to be made, more or less consciously, as to whether the vessel is to be supported, repaired and re-caulked in this state; or whether it is to be forced to settle onto a cradle (or onto keel blocks) that will restore the original line of the keel. There is no simple resolution of this problem even when the ship has to be floated into a dry-dock: opinions differed, as revealed in professional discussion that arose in connection with the design of floating dry-docks that would themselves flex in use [38].

We have even fewer sources for hauling out than for launching. Polemics about careening and the (alleged) damage that its (alleged) introduction was doing to naus of the Carreira da Índia about 1600 suggest that even very large ships were hauled out for repairs in places like Goa, which had negligible tide for the purpose of repair of large ships, and no dry-docks. One of the earliest accounts located is in fact from Madapollam on the east coast of India about 1680, in which Bowrey briefly describes the local methods employed to haul out a 1,000 ton (sic) ship. The text is as follows (Fig.9) [39]: "...their launching and hauling up the ships is after a most excellent manner, for which they are highly to be commended. I have seen...a ship of great burthen, built for the trade to Mocho in the Red Sea, and after two voyages thither she was hauled upon the western side of this river..., to the intent they might sheath and repair her. She could not be less (in my judgement) then 1,000 tuns in burthen, and they hauled her up by strength of men with good purchase...four-fold tackles, the fall of 15 or 16 inch [381-406 mm circumference] coir cable, the which are brought to two very substantial crabs, placed a little above the height they purpose to heave the ship to, and heave first at one end and then at the other 5 or 6 foot at a time, and so on until she is high enough, the dogs running upon good rollers...". Bowrey's sketch shows the hull hauled out broadside-on. In some unspecified way cradles have been placed under the hull, and are secured to the hull with ropes. In this case the bilgeways (Bowrey calls them dogs) extend across the keel, and there are packings and wedges (beds and quoins, together Bowrey's cradle) placed above them to roughly match the hull. These assemblies would be ballasted to neutral buoyancy, and moved sideways under the ship in contact with the keel until the cradles made contact - though not necessarily good contact. In fact Bowrey draws the hull inclined with only one side of the cradle actually in contact. How much faith can be placed in this level of detail is unclear: he also draws the dogs sitting clearly above the real line of the keel and unsupported bilges, despite his text description, and spanning the full breadth of the ship. The dogs are said to be single timbers at each end, 20 feet long (6.1 m) by about 20 inches (508 mm) depth. There certainly had been Indian vessels of 1,000 tons, but not of 20 feet beam - 1,000 tons burthen (previously taken on trust by this writer) would require about 40 feet (12.2 m) beam. The vessel may have been 100 tons burthen. Square holes in the dogs for the fids are virtual proof that the dogs were not single timbers at each end of the ship, but a series placed side by side, like Fernandes' vasos, producing much larger and more stable bearing areas for both the hull and the rollers. At this date it is impossible to know whether the dogs and their fids are a copy of European methods, or a local development. The text does not say so, but it is probable that boards were laid on the ground to prevent the rollers digging in - otherwise they are simply crude groundways, will not roll to reduce friction, and 1,000 tons would be very improbable with such limited bearing area. (Perfect rollers concentrate loads intensely, tending to cause self-defeating damage; imperfect rollers are little help).

The next account to consider is by Ollivier, for southern France, about 1736 [40]. His text is: "The cradle to drag vessels ashore is made in the ports where there is no tide, with three bilgeways, one of which is placed under the keel, and the other two one each side of the keel, like the bilgeways of a launching cradle. The parts of this cradle are the bilgeways, the transoms, the stretchers, the stoppings-up, the dagger planks and the chocks. There are no poppets or lashings in the cradle. The pieces of which it is composed are fastened one to the other with nails. They are assembled on a slipway, so that the underside of the cradle may be in a straight line. One gives to the upper face of the centre "bilgeway" that must touch the keel the same arc that one observes in the keel of the vessel..., and raises the stopping-up and dagger planks above the bilgeways each side as the figure and form of the vessel demand. The cradle thus built is launched into the water; one loads it on the sides with old cannon to make it sink, and leads it just under the vessel,..., the weights with which it is loaded are removed. In the [strongly tidal French Atlantic] ports of the ocean, low tide allows the cradle for the vessel...to be set up on the avant-cale. It is composed of two bilgeways like a launching cradle. One puts the stoppings-up on these bilgeways, supported on chocks, such as the figure of the vessel demands, or [sic: and] one fastens these pieces together with transoms and stretchers, and offers up the vessel on the cradle at high tide, so that when the tide falls, the vessel is supported on the cradle".

A series of descriptions of machines for hauling ships ashore are given in Machines et Inventions, Tome II by the Academie Royale des Sciences for 1702-3. One of these is for the method actually used at Brest, and most of the other ports of France, according to its author, Blanchart. Du Mé proposed a series of fixed rollers. The force was still to be applied with a prodigious assembly of very heavy hawsers and thirteen massive pulley blocks, secured between eight half-buried anchors for the standing parts of the tackle and powerful geared capstans for the falls. The untarred hawsers were of 9 and 15 pouces circumference (77.5 and 129 mm diameter). The sheaves of the blocks were to have two and a half pieds (812 mm) diameter, the multiple blocks using only the smaller hawsers. There were four geared capstans in this arrangement. Du Mé continues: "Thus when one wishes to drag a vessel ashore, the ordinary apparatus supposed to be made, that is to say the vessel established upon its cradle,...because there are many inconveniences and risks in this method of hauling vessels...One cannot pay too much attention to making the vessel bear upon its cradle, where it is always in danger of turning over...The different shocks caused by the work, and the different turns that the vessel is given in this situation in relation to its weight, often take it on one side (de faux côtés), altering its construction absolutely....If a hawser should break, it can result in many accidents, both to the vessel and to the workmen...Nevertheless these inconveniences have no remedy, since an almost similar manoeuvre is made use of every day for the same purpose".

In some ways the machine actually in use at Brest [Fig.10] is the most interesting account, containing as it does perhaps the earliest extant calculation of the actual forces involved. This related the number of men (36 to each capstan, 216 in all, each pushing with a force of 25 livres (12.5 kgs)) and their places on the capstan bars, to the force applied to each part of the apparatus, tracing it back to a total force applied to the cradle, and allowing for the slope of the slipway (1 in 25 here), the weight of the vessel that would be in equilibrium - the capacity of the machine. The critical fault in the process is that no friction is allowed for: not in the axle of the capstan, nor in each sheave of each pulley, nor between the cradle and the slipway. The correct answer is certainly not over 5,000 tonnes as stated but nearer 900 tonnes, which probably does correspond with the launch weight of large ships of the period. Nonetheless, this is the first such published estimate located, and the fact that Academicians were concerning themselves with such matters, when the whole business of shipbuilding and launching was not normally a literary or socially prestigious activity, is an indication of the practical problems States faced in operating large ships.

It is unusual to find information about the construction of slipways, or even about the range of techniques that might have been used in their construction. Thus Fernandes provides us with some details of the grade in 1616, but limited to a length that is no more than that of the hull: it is not clear what happened in the gap between the building area and the point at or below the low water mark where a large ship could be floated free of a cradle. Ollivier refers to the avant-cale that was required for launching in that zone, but his detailed text on the subject, if ever written, is lost. In the Mediterranean, for example, the only alternative even for quite small vessels would be to drag them over the natural sea-bed.

A number of techniques were certainly available in principle for constructing the slipway beyond the low water mark where necessary. Caisson construction is an ancient skill, with considerable remains reported and methods reconstructed especially from Caesarea, though this was actually for the construction of harbour moles. Huge timber boxes were constructed, floated into position, sunk and filled with concrete or stone and hydraulic mortar [41]. They are described for harbour works by Vitruvius. Classical methods could equally place individual stone blocks of at least nine tonnes weight.

The first text that mentions the matter seems to be Diderot's Encyclopédie of 1751, which has an entry for cale that is worth giving, as an excellent summary of the problem: "...prepared to a gentle slope, and extending just into the sea, to drag the vessels ashore when there is a question of repairing them. The principal inconvenience that is found in slips is that the vessel is in danger of falling onto its side when it is dragged onto the slip, or when it is returned to the water; and when the ship rests on the slip it can only be supported by the bilgeways, which cannot go from one end of the vessel to the other because of the rising forms of the bow and stern, only one part being supported, while the bow and stern that are not supported by anything suffer greatly. Besides, the slip being narrower than the vessel, it cannot be shored from one end to the other. These inconveniences are not met with in the dry-dock. So that a slip be as perfect as it may, it is necessary that the bottom is made very solid, and extremely smooth, maintaining a gentle slope equal to about 6 to 8 lignes per pied [1 in 18 to 24], so that it becomes extremely long, and can have a length of around 600 pieds [195 m], by 25 to 30 pieds [8-10 m] in breadth. It is necessary that it extends under the water to the extent that it has at least 21 pieds [6.8 m] of water at the end, so that a ship may bear entirely on the slip, and that the keel touches from one end to the other at the same moment; because a vessel of which one part touches and another is afloat suffers greatly. To render the bottom of the slip solid, it is made of great masonry-filled caissons (caisses maçonnées), for which it is necessary to pay great attention to placing them in such manner that the level of the slope is carried forward well. The coffer for the end that is advanced furthest under water is very difficult to sink. A grillage of timber called an échelle is put on this base, that serves to make the vessel slide, and ribbands (coulisses) are set up there so as to drag the vessel straight and stop it veering. Several capstans are used to drag the vessel onto the slip, and a timber construction that is called a berceau. There are necessary for a slip one grillage, three cradles, one for large, one for average and one for small vessels, and several capstans".

Nonetheless, there must be a suspicion that formerly the slipways really were stopped short at the low water mark, and shipbuilding sites selected with the right slope and firm ground and no other preliminaries. The apparent absence of texts on the matter, and the drawings of Fernandes, Gaztañeta and Colbert might support this slightly surprising conclusion. This would go some way to explaining why ships had to be dragged afloat.

This ship was the great nau of its age, intended to steam non-stop to India and beyond. The reasons for referring to this seeming anachronism are several. It was a national event, for the greatest ship ever seen, with an iron hull weighing 12,000 tons at launch; it was a mixed success (for reasons which are still the subject of bitter partisan dispute), and this combination makes it probably the best-recorded launch in history. It was extensively photographed [42]. The components of the cradle (some of which can be traced back to those for India naus) and ways were elaborately tested in advance. The details of the launch process and the results of the preliminary tests are reported in Brunel's biography [43].

There are thus important points recorded that are of relevance in understanding why much smaller ships could become stuck upon the ways. When timber ways, however carefully prepared were loaded (to perhaps several tonnes per square foot), and slid over each other, three things happened. Firstly there was a force resisting first movement of the two surfaces to be overcome, rather greater than sliding friction: stiction. Secondly, the carefully applied grease was forced out from the sliding surface, and unless the ship attained sufficient momentum (a velocity of about 0.3 metres/sec) in the first few feet it could grind to a halt. Thirdly, any imperfections in the surfaces and wild grain in the timber could lead to the grains so interlocking that if a ship stuck on the ways for this reason and the bilge and sliding ways were then cut out, they could only be separated with great difficulty: they had become "wood-bound". Thence Steel's insistence on careful preparation of the surfaces, with no projections to catch, and no weak spots in the ways or their foundations, to cause uneven loading. Brunel was well aware of these risks, and preferred to try the use of iron surfaces. The conclusion was that the critical feature was initial lubrication, for the cradle to start moving; which corresponds to the regular provision of drivers for the first impulse in old methods.

Even at the height of Victorian engineering confidence, then, the whole process was perceived as exceptionally difficult. Many of the leading Engineers of the day gathered to watch and learn, fully aware of the significance of the events. The increasing size of India naus in Portugal from 1500 onwards must have presented just the same challenges to their builders.

Conclusion.

The paper has shown that launching a large ship was a complex and difficult operation at any time, just as subsequent repair of that ship below the waterline was a major problem either in a home port, or at the far side of the world. We have seen how galleys and small coasters were relatively easy to manoeuvre on rollers and greased planks, but also that when during the sixteenth century ships were more commonly built to a thousand tons and more, they had become a major problem, with no ready answer available to the shipbuilder. Such ships caused great difficulties in launching them, and the process often took many days.

It seems possible that the importance of continuous bilgeways and of longitudinal sliding planks above the groundways had not been recognised. In Southern Europe the articulated vasos of the old galley tradition were binding on the transverse timbers of the slipway until the mid-seventeenth century. The timbers tended to squeeze out the grease and to tear each other to bits and become wood-bound, rather than slide the ship freely in its cradle. It is only in the late eighteenth century that we can be certain that sliding planks were commonly used above the groundways for large ships, even in Northern Europe.

Nonetheless, there was a steady improvement in the design of cradles during the seventeenth century, reducing the bulk of the cradle, and increasing its efficiency. This extended to methods of getting ships ashore for repair, and may perhaps be reflected in the change from launching ships bow-first to launching them stern-first, which was a major advance. There is limited, but not consistent evidence that Portuguese shipbuilders adopted stern-first launching during the sixteenth century, the first textual record of which is from Witsen in 1691.

The other striking aspect is the absence of early accounts of adequate slipways into deep water: there is no description until the mid-eighteenth century. Several sources have drawings that almost refute their existence, as we have seen. This too must have been critical to the launching process, and to the need to drag ships afloat.

The inheritance of methods suitable for rope and muscle, and of difficulties anticipated, extended to influence the launching of the early leviathans of the age of iron and steam. The problem for the builders of India naus should not be underestimated.

This brief survey of the sources lacks contemporary evidence to answer some of the riddles: there is considerable scope for more research into such practical matters. Not least that of when and under what circumstances large ships were first launched stern-first in Portugal.