Life at Sea Part Two banner

In Dumas’ The Count of Monte Cristo, the sailor Penelon tells the story of a crisis at sea he had survived. His ship had pitched heavily for twelve hours, scudding under bare poles in a gale, and finally sprung a leak. Penelon continues:

“All hands to the pumps,” I shouted; but it was too late, and it seemed the more we pumped the more came in. “Ah,” said I, after four hours’ work, “since we are sinking, let us sink; we can die but once.”

Hearing this, Penelon’s captain fetched a brace of pistols, and said, “I will blow the brains out of the first man who leaves the pump.”

Plainly, there is dramatic potential when a ship is in danger of foundering. And, inevitably, a ship takes on water.  Waves crash over the bulwarks or surge through open gunports, rain falls on deck, enemy gunfire, icebergs or submerged rocks may pierce the hull, and the hull itself leaks. The increased weight of the ship, attributable to the unwanted water, reduces the reserve (net) buoyancy and, if the process is not arrested, ultimately the ship sinks.

The risk of foundering can be reduced by good ship design, but the time will come when the ship needs a good pump. Or more than one. Let us see how these concerns were addressed in the 17th century, and thereafter.



Ship Design

Freeboard. The ship’s vulnerability to taking on water by wave action depends in part on its freeboard.  That’s the distance from the waterline to the height of the uppermost continuous deck exposed to the weather and sea, which has permanent means of closing (i.e., hatches), and below which the sides of the ship are fitted with permanent means of watertight closure (assuming that all lower decks are fully enclosed). If that deck varies in height (a ship which is higher at the ends than at the middle is said to have sheer), then it’s the lowest point that counts.

Freeboard is measured assuming the ship is upright, but in fact the ship will pitch and roll, lowering the distance to the waterline here and raising it there. How much it does so will depend on the ship’s initial stability and the conditions to which it is exposed.

Lloyd’s at one time required that insured ships have three inches in freeboard for each foot of depth of hold (Taylor, Muckle’s Naval Architecture (2013) 46). Later, complex tables were introduced which considered ship type, length, depth of hold, the coefficient of fineness, the sheer, and enclosed superstructures. An iron or steel sailing vessel with a length of 300 feet, depth of 30 feet, normal sheer, no enclosed superstructures and a coefficient of fineness of 0.68 would have a required freeboard in the summer North Atlantic of seven feet (Owen, The Tonnage and Freeboard of Merchant Ships (1906) 21-48).

After WW II, it was recommended that freeboard for warships equal 1.1 times the square root of the length in feet (Brown, The Tonnage and Freeboard of Merchant Ships (1906) 214; see also Brown, The Grand Fleet: Warship Design and Development 1906-1922), probably taking into account the relationship of maximum speed to length and the resulting chance of taking green sea over the bow, but that high a freeboard is probably appropriate only with engine-powered ships and steel construction.

LaS-P2mryrsWhen a warship opens its gunports, water may surge in through the open ports. Insufficient gunport freeboard (16 inches!) paved the way for the loss of the Mary Rose in 1545 (Sephton, Sovereign of the Seas). Reviewing the original 1664 plan of the Warspite, Charles II insisted that the freeboard of the lower deck gunports (i.e., the distance from the waterline to the lower sill) be increased to 4.5 feet (Winfeld, British Warships in the Age of Sail 1603-1714 (2010) 54).

Heavy seas could reduce the disparity in fighting power between an eighteenth-century frigate and a two-decker, as the frigate didn’t have guns on its enclosed decks and the freeboard of its main open deck was likely to be greater than that of the larger ship’s lower deck gunports. For example, the mid-century French Embuscade had a freeboard of about eight feet (Sadler, Blood on the Wave: Scottish Sea Battles (2012)). Freeboard on early nineteenth-century British frigates was usually six to nine feet (Gardiner, Frigates of the Napoleonic Wars (2006) 143).

Stability. A ship with low stability will tend to pitch and roll to a greater extent, and thus is more likely to take on water. Stability is actually a complex concept. For example, while increasing freeboard increases reserve buoyancy, it also raises the center of gravity (all else being equal) and thus reduces initial stability. And if freeboard is increased at the expense of beam (to keep underdeck volume constant), that, too, reduces initial stability.

One of the important parameters of stability is the metacentric height (GM), the distance between the center of gravity (G) and the metacenter (M), the point where the vertical line through the center of buoyancy (B) on a heeled ship crosses the vertical line through B on the unheeled ship. While an increase in GM increases the righting moment (the moment acting to correct the heel), it also reduces the natural period of roll, and a ship with too short a roll period can be uncomfortable to ride and also vulnerable to dismasting.

A more detailed discussion of ship design relative to stability is outside the scope of this article.

Bulwarks. A further defense against wave action is the bulwark, essentially a wall on an open deck. Magoun (The Frigate Constitution and Other Historic Ships 51) says that the Mayflower, already an old ship when chartered in 1620, had solid bulwarks. Despite those bulwarks, it was a wet ship.

Contemporary illustrations show that many ships in period didn’t have a solid bulwark, just a rail supported with stanchions; this might give a crewman something to hang on to but wouldn’t keep out water. While sometimes canvas was hung over the rails, this would just keep out sea spray and not green water.

Even if the ship had a bulwark, if it were armed, it might either have a relatively low bulwark so the open deck guns could fire over it, or gunports cut into a higher bulwark. Normally, these open deck gunports lacked lids, so the effective height of the bulwark in terms of watertightness was the height of the lower sill.

A 1918 text states that the average height of the bulwark is 4.5 feet in sailing ships and 2.5 feet in steamers (which are not heeled over by the action of a beam wind on sails). By that time, ships had metal hulls and the bulwarks were thick plating (Holms 345).

Camber. A ship’s deck is cambered—curved so it’s higher at the centerline than at the sides—so any water runs off to the sides. Of course, one must still get the water off the deck.

Thomas Harriot’s Notes on Shipbuilding (1608) suggested that the camber (center-side height difference) should be about one-half inch for every foot of half-breadth, a 1:48 ratio (Lavery, The Colonial Merchantman: Susan Constant, 1605 (1988) 16). On the mid-seventeenth-century fluyt Zeehaen, the hull breadth was twenty-two feet but the camber on the upper and lower deck beams was ten inches, almost double the relative height. (Hoving, The Ships of Abel Tasman 127-8). Nicolaes Witsen taught that the camber of the lower deck beams should be one inch for ten feet of length (and the length was supposed to be four times the breadth) (Hoving, Nicolaes Witsen and Shipbuilding in the Dutch Golden Age 74, 250). On the Bellerophon, “the camber was six inches on the gun deck and an inch less for each deck above” (Pope, Life in Nelson’s Navy 39).

Scuppers. The bulwarks that keep the smaller waves out also trap the water left behind by waves that crested the bulwark  Hence ships are equipped with scuppers, deck level openings through which the water can drain off. The scuppers are usually on the main open deck (the weather deck), which is above the waterline. However, they can be lower, and channels conduct the water from the higher deck to the one the scuppers are on.

That of course leads logically to the question, what keeps the water from entering by way of the scuppers? On the Mary Rose, a leather flap was nailed over the scupper hole. This acted as a one-way valve (McElvogue, Tudor Warship Mary Rose 21).

The scuppers must be dimensioned to take into account the volume of water that can be trapped by the bulwarks. Holms (345) advises that the combined area of these freeing ports should not be smaller than 10% of the area of the bulwarks.

Bulkheads.  These divide the ship into watertight compartments. Thus, if there is water leakage into one of them, the maximum loss of buoyancy is the volume of the affected compartment.  Bulkheads may be transverse or longitudinal, and made of wood or iron.

At the time of the Ring of Fire, bulkheads had been used on Chinese and Japanese junks for centuries. There is some dispute as to whether these bulkheads were in fact intended to be watertight, as they had limber holes at the bottom, but I agree with Chinese scholars that these were intended to facilitate washing cabins (the boat could be trimmed by the stern, so the water would drain sternward and there be pumped out) and at sea the limber holes would be plugged (Cai Wei, et al., “Watertight bulkheads and limber holes in Ancient Chinese Boats” in Jun Kimura, Thematic Studies in East Asian Maritime Archeology, 2010,

In 1787, Franklin proposed that the holds of packet ships be “divided into separate apartments, after the Chinese manner, and each of these apartments caulked tight so as to keep out water.” In 1795, Bentham likewise advocated “partitions contributing to strength, and securing the ship against foundering, as practiced by the Chinese of the present day.”

Watertight bulkheads remained uncommon in the West. Most nineteenth-century sailing ships had merely a collision bulkhead. Those that had more were mostly steamers converted into windships, and the most bulkheads on any unconverted ship was four. Even steamers weren’t necessarily compartmented. In 1881-3, one hundred twenty British iron steamships were lost that “had a single compartment the filling of which would have caused the ship to founder” (Barnaby, The Protection of Iron and Steel Ships against Foundering from Injury to their Shells, including the Use of Armour, J. Iron & Steel Institute 37: 438 (1890) 445).




The rate of inflow from a hole below the waterline is usually proportional to (1) the size of the hole and (2) the square root of the depth of the hole below the waterline. If the water rises inside the hold enough to cover it, the rate of flow is proportional to the square root of the distance between the water level inside the hull and the waterline outside (Oertling n4).

It follows that as the boat fills with water, the rate of water entry slows down (assuming the water doesn’t create new holes). Thus, a point can be reached at which the rate of water removal (by bailing or pumping) equals the rate of water ingress—the ships stays afloat even though it is waterlogged. A leak in the bow was more dangerous than one in the stern because the forward movement increases water pressure at the bow.

While a cannonball could certainly create a large hole, it wasn’t that easy for enemy fire to hit a ship below the waterline. The most common leak was the result of a planking seam which had lost its caulking. The leak could be located by listening for it with an ear trumpet.

Leaks could be plugged from the inside or outside. On the inside, one could use some sort of gelatinous mixture (like tallow and coals), pieces of raw beef, oatmeal bags, sheet lead, or canvas or leather backed with oakum.  On the outside, one lowered a bag or net of oakum down over the leak, which then sucked in the oakum. Shot holes were usually closed by driving a canvas covered wooden plug into the hole (Oertling 7).



Bailing and Scooping

With ten-liter buckets, “one man can lift about 15 buckets per minute or 300 cfh to a height of 3.3 feet” (Wood, Pumps and Water Lifters for Rural Development (1977) 29). Water may also be thrown by using a semi-enclosed shovel, with about the same rate of water transfer (41). A trick that increases the rate of scooping is to attach the shovel by a rope to a tripod so you get a “pendulum assist.”


Ship Pumps

If the deck water doesn’t escape by way of the scuppers, then it will eventually drain down to the bilge. Water entering by way of leaks or shot holes will do the same.

It was not very practical to use a bucket brigade to carry water all the way up to the weather deck in order to dump it out. In theory, it would have been possible to place a bucket on a rope and use a pulley to pull it up a great height. Moreover, one could use two buckets, attach to opposite ends of the same rope, and replace the simple pulley with a cranked roller for increased mechanical advantage. However, I am not aware of the use of this ancient device on shipboard. Instead, pumps were used.

The pump drew water up through a pump tube, whose length was dictated by the depth of the hull. The pump and pump tube were inside a compartment called the pump well.

Encyclopedia Britannica 11th Edition (EB11) /”Pump” says that the simplest type of pump used to move a liquid is a plunger pump, characterized by a piston moving in a cylinder, and various valves. The plunger pump type is subdivided into suction pumps and force pumps. Suction pumps will be discussed in more detail shortly.

Force pumps have a solid piston (there is a valve inside the piston of a suction pump) and the outlet is below the piston (rather than above it as on the suction pump).  Liquid rises on the upstroke, and is forced into the outlet by the downstroke (with the inlet closed by a check valve).

EB11/”Hydraulics” briefly mentions lift, bucket and plunger, diaphragm, chain-and-bucket, and scoop wheel pumps, but focuses on centrifugal pumps (see below).

A more cogent classification system appears in the modern Encyclopedia Britannica (2002 DVD, essentially corresponding to the 1999 print edition).  Water pumps are classified as to whether they work by volumetric displacement or by adding kinetic energy. The displacement pumps are classified as being reciprocating (piston, plunger, diaphragm, etc.) or rotary (gear, lobe, screw, vane, or cam). Reciprocating pumps can be single- or double-acting, the latter pumping on both strokes. Kinetic pumps are classified as centrifugal (radial, axial, mixed flow) or regenerative.

In the 1630s, there were three basic types of shipboard pumps: burr, common (suction), and chain pumps.

Burr Pump. Described by Agricola in 1556, it had a vertical pole that moved up and down inside the pump tube. On the lower end, the pole was thickened (the burr) and to this was attached a leather cone (“shoe”). Strips of leather ran from the base of the cone to an anchor point above the base.  On the downstroke, the cone closed and water entered the pump tube. On the upstroke, the cone opened and water was carried upward. (Agricola, De re Metallica (Hoover transl., 1912) 176; Ewbank, A Descriptive and Historical Account of Hydraulic and Other Machines for Raising Water (1876) 214; Oertling 24ff).

The burr pump was also equipped with a foot valve at the bottom of the pump tube. Oertling doesn’t say much about it, but a foot-valve is a one-way valve used to keep the pump primed (i.e., filled with liquid).

By the seventeenth century, the burr pump generally was “no longer in use on English ships, but could be found on Dutch and Flemish ships.” However, it was occasionally seen, in modified form, as late as 1860 (Oertling 29).

Oertling says that it was difficult to service; the entire pump tube had to be lifted off its base. On the other hand, according to Boteler (1634) and Manwayring (1644), it drew up “far more water and was less labor intensive than the common pump at that time” (Oertling 30).

Common (suction) pump. First described in 1433, and used in mines and on ships in the sixteenth century, if not earlier. It features a fixed lower valve, an air-tight piston that moves up and down in the box, and an upper valve inside the piston. On the upstroke, a vacuum is created between the piston and the lower valve, drawing water up through the latter. On the downstroke, the drawn water is forced up through the upper valve and ultimately to a spout.

The theoretical limit for raising water by suction is about thirty feet (assuming barometric pressure of air is 30 inches mercury). Because of friction, the practical limit is 28 feet, and this is measured from the surface of the water to the closing member of the upper valve.  The pump was typically placed near the center of the pump tube to reduce the critical distance.

The lower valve was a lift check valve, that is it had a central hole closed by a vertically movable leather claque held down by a valve weight. Water pressure could lift the claque and weight, opening the valve, but a guide maintained its alignment. It was equipped with a staple so that it could be fished up with a hook at one end of a long pole, for repair.

The upper valve-cum-piston had a wood body, with a slot to receive a wooden shaft (“spear”) at the upper end and a check valve covered with a leather gasket at the piston end. This could be a lift check valve, but one recovered from the Machault (1757) had a hinged claque instead. Both lower and upper valves were usually made of elm or ash.

The spear was pivotably connected at its upper end, above the top of the pump tube, to one end of a lever (“brake”), whose fulcrum was provided by a “cheek” that curved away from the top of the pump tube.

Common pumps could be connected in parallel (two piston cylinders connected by a T to a pump tube communicating with the bilge, as in Dodgeson’s 1799 pump) or in series (two pistons in a single cylinder, as in Taylor’s 1780 pump).

On the Taylor pump, the shaft of the lower piston ran through the upper piston and was connected above it by a jog section to one side of a cog wheel, and the upper piston was connected at its periphery to a shaft that in turn was connected to the other side of the same cog wheel.  Thus, when one piston went down, the other came up. the cog wheel was connected to a double action pump brake or drum. It would thus produce twice as much water as a single action common pump of the same bore and piston stroke (Oertling 64ff, Ewbank 226).

Chain pump. Chain pumps have a curious history; while known in Roman Europe, the concept was lost, and was reintroduced via contact with the Tartars of eastern Europe in the fifteenth century. It was then used to drain mines, whether as a result of technology transfer from the European mining industry, or from the Chinese in the sixteenth century.  Raleigh reports the introduction of the chain pump in the late sixteenth century to the British navy (Oertling 75) and it is also described by Manwayring (1644) and Boteler (1634).  Dampier, in the late seventeenth century, had both a chain pump and a common hand pump; in the mid-nineteenth century, British warships carried four chain pumps and three common pumps (Ewbank 154).

The chain pump features an endless chain bearing circular discs (“burrs”) through a vertical tube, open at the top and bottom, the latter being immersed. The chain runs around two sprocketed wheels, a drive wheel on top and a guide wheel on bottom. It enters the water, passes around a submerged guide wheel, and moves upward, the discs entrapping water when they enter the bottom of the tube. The water is carried by the discs up to the top of the tube, where it passes into a discharge channel, and the chain passes around the drive wheel and descends back to the bilge. Originally, the drive wheel was a solid wooden wheel with iron sprockets to engage the chain, and a crank attached to the shaft. The links were cast iron, and round or S-shaped. The disks were wooden.

The chain pump was able to move more water and was easier to work than the common pump, but it required a large crew, and the disks wore out quickly. It was used mostly on large warships (Oertling 80).

One problem with the old chain pump was that the weight of the water pressing down on the discs would tend to cause the chain to slide back on the sprocket wheel. The links were not well united and often broke. In addition, there was a lot of friction between the chain and the wheel, which, I imagine, increased the effort necessary to lift a given quantity of water. (Edinburgh Encyclopedia/Pump 202).

In 1764-68, Cole and Bentinck tested a new chain pump design. It wasn’t officially adopted until 1774, after further modifications were made (Oertling 78).

Cole, in British Patent 911, issued December 16, 1768, just mentions the ease of repair and not how it was achieved. Apparently, “every other link was formed of two plates of iron, whose ends lapped over those of a single one, and secured by a bolt at each end” (Ewbank 155).  The chain links were cast to the same size, and were therefore interchangeable, as were the link pins that connected the links (Oertling 93).

The links were designed so that they could be undone and a worn link replaced easily. Ewbank’s description of this is a bit difficult to follow, but Oertling (Fig. 25) has described the chain assembly from the HMS Charon‘s (sunk 1781) chain pump. In essence, the link pin has a slot near one end, and an L-shaped cotter key is inserted into the slot. Thus, to unlink, just pull out the cotter key and then the link pin. In one experiment, the chain was deliberately broken and dropped in the well; it took just two-and-a-half minutes to retrieve it, repair it, and resume pumping (Nicholson, The Operative Mechanic (1825) 268).

The burr (“saucer”) was positioned every fourth single link and it was composed of two plates of cast iron with leather in-between. The leather plate was of the same diameter as the bore of the pump tube, and the flanking metal plates were slightly smaller to minimize friction (Cole and Bentinck, British Patent 982, issued Jan. 17, 1771). Ewbank says that even the leather doesn’t actually have to touch the wall of the tube.

The drive wheel, instead of being a simple sprocket wheel, took the form of two metal (brass?) discs eight inches apart on a common axle, further united by peripheral (iron?) bolts parallel to the wheel axis—essentially a cage gear.  (The 982 patent likened it to the “skeleton of a drum.”) The links of the chain had hooks that engaged these bolts (teeth) (Edinburgh Encyclopedia; Nicholson 268) .

With four men at the crank, the Cole-Bentinck chain pump discharged one ton of water in 43.5 seconds, versus 83 (Oertling 78; Ewbank 155 says 55) seconds for the old design. This pump is probably not described in Grantville literature, but could be invented independently.

The cast iron links were replaced with brass ones in the early nineteenth century (Oertling), and still later the lower wheel was replaced with a curved metal tube, to reduce friction (Edinburgh Encyclopedia).

Wood (70) says that the lower pipe end of a chain pump is “usually flared to facilitate entry of the discs into the pipe” but I haven’t seen reference to this feature on ship pumps.

In a 1956 study, four men operating a chain pump with a four-inch pipe were able to achieve 40 cubic feet/hour discharge over a twenty-foot lift, 72 cfh over a ten-foot lift, and 110 cfh over a five-foot lift (72).

Chain bucket pump. I am not aware of any shipboard use, but this device (also called a Persian wheel) replaces the disks moving through an enclosed tube with individual buckets. They empty at the top of the movement into a discharge trough. The drive wheel is a portgarland, that is, it has projections on the rim, parallel to the shaft, to catch the buckets. The 1956 study (75) showed it to be superior to the simple chain pump, with discharge of 395 cfh over a twenty-foot lift, 580 cfh over ten-foot lift, and 760 cfh over a five-foot lift.

My guess is that the reason that this was not used on shipboard is that it is traditionally a large structure, with a wheel that is man-height or larger, and driven by animal power via a right angle drive (Yannopoulos, Evolution of Water Lifting Devices (Pumps) over the Centuries Worldwide, Water, 7:5031-5060 (2015)).

Centrifugal Pumps. These weren’t used as of the Ring of Fire, but are probably the most important modern bilge pump type.  These have a wheel with curved vanes (“impeller”) enclosed in a chamber. Water enters at the center of the chamber and spirals out under the influence of the rotating impeller.

Euler discussed its theory in 1754, and some sources say it was invented by Jordan (1680) or Papin (1689). There was a successful centrifugal pump design introduced in 1818 (“Massachusetts pump”) but it had straight vanes, and curved vanes proved much more efficient. (Greene, Pumping Machinery (1911) 43ff). The 1851 “Appold” centrifugal pump, with curved vanes, “raised continuously a volume of water equal to 1400 times its own capacity per minute.” A further innovation was the “whirlpool” zone suggested by Professor Thomson, a free vortex space surrounding the wheel (EB11/Hydraulics).

They are not self-priming, and thus must be sitting in water in order to pump it. In theory the impeller could be rotated manually by a crank, drum, or capstan. However, when they were introduced, steam power was already available.

The USS Monitor (whose freeboard was only eighteen inches (Tucker, American Civil War: The Definitive Encyclopedia and Document Collection (2013) 1312) had a steam-powered centrifugal pump capable of moving 23,000 gallons per minute, but it wasn’t enough to save it from sinking in 1862; its coal was wet which reduced effective steam power (Wikipedia).

Initially, they were driven by gearing from the main engine, but later these pumps were driven by auxiliary engines. If there was a long vertical shaft from the engine to the impeller below, they could be worked even if part of the hold was flooded. On the Inflexible, the pump engine was high enough so the pump could be worked even with twelve feet of water in the engine room (Smith, A Short History of Naval and Marine Engineering (2013) 208).

In 1961, Charmonman improvised an axial flow pump by encasing the propeller of a Thai-style outboard motor in a cylinder (Wood 112).

Diaphragm Pumps. These weren’t used on ships as of the Ring of Fire, but are sometimes used nowadays as backup bilge pumps. They have the advantage of being self-priming. Like a piston pump, they vary the volume inside the pump chamber. However, they accomplish this by moving a flexible diaphragm in the side of the chamber, rather than by moving a piston.

Pumps: Motive Power.  Generally speaking, seventeenth-century shipboard pumps were human-powered, with sailors pulling down a lever, turning a crank, or pulling on a rope wrapped around a drum. It should be noted that “during short time periods (10-15 minutes), the legs can develop about 0.25 hp while the arms can only provide about 0.10 hp. Over a sustained period (say five hours), a grown man is capable of 0.06-.08 hp (Wood 122).

That said, in the Netherlands and Great Britain, windmills were used to operate pumps used to drain land for agricultural use. (At least in China, seawater was also pumped onto land for salt extraction.)

LaS-P2wndmllIn the nineteenth century, Norwegian and Swedish ships were routinely equipped with wind pumps (Leslie, A Sea-Painter’s Log (1886) 52). However, they were less common in other merchants marine (even Dutch!). Wind pumps were used on ice barges in upstate New York, and one was rigged up on the Henry Woolley in 1871 after it sprang a leak (“A Useful Invention on Ship-Board,” West Coast Times, Issue 1633, p. 2 (Dec. 9, 1871)).

An 1876 British writer estimated that the cost of the wind pump for a vessel of 800 tons would be about 40 pounds. He assumed that it would have sails six feet long, fixed to a revolving head mounted on a bipod mast (Wade, “Windmill Pumps” (Letter), Nautical Magazine 45: 1027 (1876) 1028). The wind pump frees the crew from pumping duties, but it doesn’t work in a calm and also takes up deck space.

It is conceivable that a paddle wheel (undershot) or propeller could be used to power a pump on a sailing ship. The ship would be impelled forward by the wind, causing water to pass the paddle wheel or propeller and turn it. I would suspect that this would be less efficient than a wind pump, and would increase drag, but the deck space would be unaffected.

The other major nineteenth-century source of motive power for a pump was the steam engine. Strictly speaking, devices which used steam or heated air to displace water were developed by Heron of Alexandria (1st century AD), Giovanni Battista della Porta (1601), Jerónimo de Ayanz y Beaumont (1606), and Salomon de Caus (1615). However, here we speak of the use of steam as motive power (engine) for a drive wheel that drives a piston or chain pump. The steam pump, like the wind pump, was a labor-saving device, but unlike it, was not dependent on the wind. Of course it needed fuel to operate, and steam engines were finicky enough so that the bilge would also be equipped with a hand pump.

Note that mechanical linkages can convert reciprocal motion to rotary motion, or vice versa.

One interesting emergency expedient I found reference to was to use wave action to operate the pump. Captain Leslie of the George and Susan reported fixing a spar aloft, with one end over the spear of the pump and the other projecting over the stern. At each end he mounted a pulley, and ran a rope over the pulleys from the spear of the pump to a counterweight (a 110-gallon cask holding 60-70 gallons water, i.e., half-full) at the stern end of the rope. Supposedly, when a wave rose the butt-end of the cask, the spear was depressed, and when the wave retired, the spear was raised (Nicholson 269). It seems to me that for this to work, there would have to be a downstroke bias on the pump, that is, without an upward pull on the spear, gravity would be stronger than friction and the spear would descend. If so, the cask could be weighted to just balance the spear when the water was at a neutral height. When the wave lifted the weight, the rope would slacken and the forces on the spear would no longer be in balance, it would fall.  When the wave dropped, the counterweight would drop, thanks to gravity, and through the rope exert a tensile force on the spear, pulling it back up.

I have also found US patents (ex. Delaney, USP 3120212) dealing with wave-operated pumps—generally speaking, a float is connected to one end of a rocker arm, and the spear of a piston pump to the other—but I don’t know whether any of these have been put into practice.

Salomon (Solomon) de Caus (Caux) (1576-1626) used solar heat to expand air that in turn powered a water pump. The modern Rao solar-thermal pump uses the heat of sunlight to vaporize a working fluid like pentane at 35-40oC. The water to be pumped enters a water chamber through a non-return valve. The vapor enters the water chamber and displaces it, forcing it up a discharge pipe. At night, the vapor condenses and flows back to the flash tank. With a solar collector area of 250 square feet, and a thirty-foot lift, Rao reported a discharge of 880 cubic feet/day (Wood 97ff). Since it may take several hours of daylight to bring the working fluid to vaporization temperature, and the pumping only occurs during daylight (say 10 AM to 4 PM), this intellectually interesting system isn’t likely to work on shipboard. But that doesn’t mean someone won’t try to build something similar!

Pump tubes. While on ships, the bore of the pump tube was open on the bottom rather than plugged, the heel of the tube was seated in a hole cut in the mast step (the structure in which a mast is seated) or in the floor timbers. That would block the bore, so channels were cut through the wall of the tube at its heel to let water in. Bilge water being unpleasant from a sensory standpoint, a conduit (“dale”) was used to guide it from the top of the pump tube to a scupper at the side of the ship, rather than just spilling it out on deck (Oertling 41).

LaStmpstDebris from the bilge could be sucked up the bore and gum up a valve. The debris could be garbage, cargo, or ship stores that became wet and migrated into the bilge. The pumps of the Sea Venture (whose 1609 voyage inspired The Tempest) were clogged by biscuit fragments, and the HMS Centaur (1782) was lost when rising water caused its load of coal to infiltrate the pumps (46). To prevent this, lead, copper, or tin sieves were installed at the lower end of the tube (43).  Of course, the sieves would need to be cleaned from time to time.

Materials. In the 1630s, the tube was usually made of wood, most often elm but sometimes larch, beech, or alder. A tree with a straight, knot- and branch-free trunk was found and cut. The center was then bored out with a hand auger. Alternatively, a tube could be constructed from the hollowed halves of a log, or with strapped and caulked planks (somewhat like making barrel from staves).

It was of course possible to use metal instead, and lead tubes had been used in some urban water supply systems going back to Roman times.  Agricola (1556) suggested that valves be made of iron, copper, or brass, and lead ones may have been used on an early sixteenth-century wreck (Oertling 48).

Metal tubes were made at the time of RoF by taking a sheet whose width matched the desired circumference, rolling it so the edges met, and then soldering. Later, alternative processes were developed. One was to cast a section in a mold, draw the finished tube partway out, and then pour another section to join with the first (Oertling 56-7).

Metal tubes and pump parts were more durable than wooden one, so the objection to them related to cost. Lead use became significant in the early eighteenth century, copper and bronze in the late eighteenth century, and iron in the nineteenth century (60, 62, 72).

Pump utility. Besides being used to pump out bilge water, pumps could be used to distribute seawater for washing and firefighting. Common pumps were of limited value for firefighting, as their pressure was limited to the head pressure (weight of the column of water).


Bilge Alarm

If the ship is taking on water rapidly, the progressive reduction in freeboard will be obvious and “all hands to pumps” might well be the command.  But a small leak might go undetected until a sailor has reason to descend to the orlop deck and finds himself knee-deep. For example, in the case of the wreck of the Protector in the Bay of Bengal in 1838, while the ship was sailing under bare poles in a gale near the reefs off the mouth of the river Hughly, a midshipman sent below for grog returned hurriedly to report that the hold was half full of water (“Narrative of the Wreck of the Ship ‘Protector'”, The Pilot, or Sailors Magazine 341 (Nov 1839)).

Once the presence of water was detected, the pump well could be “sounded” to determine how much water was in the hull, and this could be monitored periodically to determine whether the pump was holding its own with the leak. On one of Cook’s voyages, at the change of shift, the new man inadvertently took the sounding at a different point, making it seem as though the leak had gained 16 or 18 inches in a short period of time and inspiring the men at the pump to “redouble their vigor” (Lamb, Exploration and Exchange: A South Seas Anthology, 1680-1900 (2000) 84).

Modern ships are equipped with bilge water level detectors that trigger an alarm or even turn on pumps automatically.

The simplest sensor design is probably a float that activates a switch when it climbs to a set point. While in modern systems, this is an electric switch, it is also possible to use mechanical linkages to create a visible or audible signal. Bilge water alarms are described in

Knight’s 1884 American Mechanical Dictionary (1:281).


Water Ballast and Trim Tanks

The lower the center of gravity of the ship, the better its stability (although if too low, the ship may become too stiff, i.e., roll frequently and violently). Stability is a particularly acute problem for warships, since guns are more effectively if mounted high. But all ships find it advantageous to carry ballast—essentially, heavy materials such as lead or iron—deep in the hold in order to lower the center of gravity.

Merchant ships that carried heavy cargos in one direction and light cargos in the other had to take on ballast after discharging the heavy cargo and then dump the temporary ballast when replacing the light cargo with a new heavy load.

This was a particular problem for colliers (coal carriers). Their owners didn’t like having to pay for the one-way ballast, and the port of origin didn’t like the ballast dumps. A collier sailing from Newcastle to London with 250-400 tons of coal would have to pay one shilling per ton for the ballast in London, and another six pence per ton to put it on board. Then, back in Newcastle, it would pay one shilling per ton to the River Commissioners (a pollution charge?) and ten pence per ton for depositing it on the river side (Holmes, Ancient and Modern Ships 2:162). Another estimate related to a merchant steamer in the Mediterranean trade, carrying 200 tons of ballast. Loading and unloading it each voyage would cost 260 pounds (1877). The steamer itself cost 20,000 pounds and made four voyages a year (163).

In 1852, the SS John Bowes was equipped with some kind of “temporary appliance” for carrying water ballast—water being free and environmentally acceptable. This experiment was deemed successful, and the SS Samuel Laing (609 register tons) was built in 1854, equipped with fixed iron water ballast tanks. The ship was double-bottomed, and the tanks rested on the floor created by the top of the inner bottom (Holmes Fig. 75). The next step, taken in building the SS Rouen, was to make the tanks an integral part of the ship structure, i.e., the top of the tank was the inner bottom.

Naturally to fill and drain these water ballast tanks, pumps were necessary, but by the time they were introduced, pumps were steam-powered.

If the ship has, not a single ballast tank, but separate tanks fore and aft, then by pumping water forward or backward, the trim of the ship may be adjusted.

The principal disadvantage of water ballast is that water is less dense than iron (Sp. G. 7.87) or lead (Sp. g. 11.35). Hence, it is less efficient (on a per volume basis) at lowering the center of gravity.


Henry Scott Tuke painted a picture entitled, “All hands to the pumps!” We see five men frantically working the seesaw arms of a deck pump, with the discharge tube bringing water up from the bilge to the deck. But was it working fast enough to save the ship?

In the new time line, we are likely to have more efficient pumps at earlier times than in the old one, keeping at least some ships from foundering.

But keeping the ship dry is just one aspect of maintaining a healthy and safe environment on board. We will look at additional “ship’s systems” in part 3.


To Be Continued . . .