Friday, September 7, 2012

The Mtepes of Kenya

A dau al mtepe (source unclear). Click any image to enlarge.
Among the indigenous boats that disappeared in the 20th century was the mtepe, the world's last large sewn vessel. The last mtepe built was in 1930, and the last one in existence was wrecked in 1935 (they might have been the same vessel: my sources are unclear).

History and Usage
            Unique to the Swahili culture of east Africa, mtepes were built in the Lamu archipelago off Kenya. Evidently an old design, their origin is unknown: there are references to sewn boats in the region that date back to classical antiquity and to the earliest years of Portuguese exploration, but these might have been other designs that were similar to mtepes in some respects. The earliest unmistakable references to mtepes date from the 19th century.
Both mtepes and dau al mtepes (shown) carried very large spreads of sail and were reportedly fast sailors and surprisingly weatherly.
            Mtepes were of two types, the mtepe proper, and the dau la mtepe, which superseded it. The transition apparently occurred in the 19th century because, while there are contemporary models and documentary evidence of the earlier mtepe, the only photos seem to be of the later dau la mtepe. Both were double-ended and generally similar in design, but there were a few significant differences between the types, which we'll get to below. (I will use "mtepe" in Roman text to refer to both types, and will italicize it as mtepe to refer to the mtepe proper as distinct from the dau al mtepe.)
Model dau al mtepe (recognizable by the bowsprit) at National Marine Museum, Greenwich. The tassels are called the "camel's beard."
Mtepes varied considerably in size. Most seem to have been between 40 and 75 feet (I presume that the measurements were taken on deck, and did not include prow extensions, bowsprits or outboard rudders)  and of 12 to 50 tons burthen, but one recorded in the 19th century was 30 meters long and 186 tons. All references report them to have been very fast and able sailors, easily able to out-point settee-rigged dhows. This is surprising considering their rig, which consisted of a single square sail of palm leaf matting.
            Faza in the Lamu archipelago was the center for mtepe building: as many as 20 were being built there annually early in the 20th century. With a crew of about 20, they traded north to Zanzibar and south to Kenya's Mrima coast. Toward the end of their employment, their primary cargo was mangrove poles, but earlier cargoes included grain, pottery, cattle, salt, firewood and slaves. To collect the mangrove cargoes, the crew would punt the vessel into a swamp and cut and load the poles themselves, receiving payment from the proceeds of the voyage on a "share" basis. Oars were also used for propulsion when necessary.
Mtepe model at Science Museum, Kensington. Note large rudder, the bird-like figure on the tall prow, flags, tassels and painted decoration, as well as the thatched house aft of the mast.
Mtepes leaked copiously, so bailing was constant. It is recorded that, as a test for employment, a prospective crew member had to demonstrate his ability to lift a bailer full of water measuring 24" in diameter and 9" deep. (This works out to about 3 U.S. gallons or 25 pounds. As a test of strength, lifting it isn't impressive, especially when considering that the bailer was made of woven palm leafs and probably leaked, and that it was very wide and shallow, so unlikely to have been lifted full in any case. On the other hand, if you're bailing a vessel, lifting that weight several hundred times per hour is indeed an impressive demonstration.)
A weatherly hull, but not made for beaching without supports. Image from Hornell.
With their deep V-shaped hulls and flexible sewn construction, mtepes were not stiff enough to lie on their sides. They were never allowed to take the ground unless supported by props.

            The mtepe bears many similarities to dhows and the two might have influenced each other's development in unknown ways in ancient times. Certainly, the dau al mtepe (which I think translates roughly as "dhow-style mtepe") adopted some of the characteristics of the dhow late in the 19th century.
Sketch of a model of a mtepe at Ft. Jesus, Mombasa.
A crew of five or six carpenters, being supervised by the ship's future captain, would typically built a mtepe in two to three months, including harvesting their own timber, which was used green. Like all relatives of traditionally-built dhows, mtepes were built shell-first, on a keel, with smooth-laid planks of mangrove wood. (I hesitate to call it carvel, because that implies a construction method and certain details which do not apply here.)
            Garboards were sewn or stitched to a one-piece keel. It is not known whether the keel was rabbeted. On many dhows, the garboard butts against the side of the keel or a beveled surface at the intersection of the top and sides. Stem and sternpost were lashed to the keel only after two or three strakes were in place. There were no deadwoods or stem knees. Partial-length planks were scarfed diagonally without butt blocks.
            Planking the hull was an enormously laborious process. After holes were drilled through the planks near both edges, the planks were sewn together with three-strand coir (coconut husk cord). First, coir fiber was pounded into the seam from inside. Then, over the seam was spread a thick paste of made of pounded mangrove bark, the tannin in which preserved the coir caulking. Onto the paste were laid strips of crushed coconut husk, and then a layer of palm leafs. A sharpened piece of palm leaf was attached to one end of the cord and used as a sewing needle. Men worked in pairs, inside and outside the hole.
Each time the man inside passed the cord through the hole, he would pound it with a mallet to compress the package of caulking material beneath the stitch, then would plug the hole with a tapered plug to hold the cord in place temporarily, until his turn came for the next stitch. The man outside would wrap the cord around a short stick, place one end of it against the hull, and lean back, using the stick as a lever to tighten the cord before passing it back through the next hole. About six feet of seam length was sewn at a time, the coir being passed around and through each hole several times, both straight up and down and diagonally. Tapered plugs were then pounded into each hole from the outside.
After the the hull was completely stitched, the plugs outside the hull were cut off flush, and then the stitches themselves were cut off on the outside! This would leave the planks fastened to each other by thousands of very short pieces of cord, held in place only by the plugs pounded in from outside – a very odd way to address the vulnerability of the stitches outside the hull. This last process was omitted only near the stem and sternpost, and apparently some mtepes were built with the stitches remaining on the outside over their whole surface.
Alternate view of plank pegs. Above: oblique (through outer face of lower plank of each pair). Below: edge-set (drilled into the adjoining edges of plank pairs). The direction in which the pegs used to plug the stitch holes were installed also differs in these two views. (Source: Adams)
            Most reports indicate that after the strakes were sewn together, pegs were inserted by drilling obliquely upward into the outboard face of a plank so that the drill exited the plank on the edge and penetrated the plank above it also on its edge. Others disagree, claiming that the strakes were first edge-drilled on their matching faces, "dowel tenons" set into the holes, and the upper strake then pounded down onto the lower one. Only then were the planks through-drilled and sewn together.
            If the first method (oblique pegging) is correct, then the pegs' main function seems to be to eliminate shear between adjacent strakes. If the second method (edge-set pegs) is correct, then the pegs seem to function primarily as an assembly aid. In both cases, the stitches are primarily responsible for holding the planks tight against each other. It is not impossible that both methods were used.
Cross section of a mtepe, showing futtock frames, stringers, thwart pairs with gammon lashings, mast step, planks and keel. (Source: Adams)
Interior framing was added only after strakes were in place. Amidships, there were futtock frames which rested on floor timbers but were not attached to them. Moving outboard were half-frames, and then one-piece frames (i.e., gunwale to gunwale) in the very ends. The frames were lashed to the hull with a few light cords passed through holes in the planks. Frame spacing was reportedly wide, but no specifics are available. Five or six sets of cross-beams or thwarts were installed in pairs, upper and lower, separated by a few planks. These were notched and passed through the planks, and were gammoned to each other. In addition to transverse strength, the upper thwart of each set served as deck supports and mast partners. Two or three stringers were installed, so lightly lashed to the frames that they seem to have provided little lateral strength: perhaps they helped support and secure the cargo.
            During the southwest monsoon, when the mtepe was laid up annually, every stringer and frame was removed and every stitch was replaced. With a full crew working, the job could be done in about eight days.

            Although most reports indicate a plum mast, more recent measurements of models and photographs indicate that the mast had a forward rake of 6 to 14 degrees. The mast was supported by a forestay, two backstays, and usually one shroud per side, although some mtepes may have had a single shroud that was passed from one side to the other when tacking – a rare occurrence for tradewind sailors.
Mast step (bow to right). Note forward rake of mast, which rests against a pair of thwarts and is lashed forward to a stanchion forward of the thwarts. A stanchion beneath another pair of thwarts hold down the aft end of the step. The step typically spanned many frames, nearly the entire interior length of the ship, but it was not a keelson. (Source: Adams)
            The sail was made from many strips of plaited palm leaves, sewn together. There were two or three sets of reef points near the head of the sail. The sail was supported by a yard and a boom of delicate appearance. The yard was equipped with parrels; the boom with a simple loop of rope to perform that function. In addition to a halyard, the yard was supported by two "lift braces," which served both functions implied by that name. As I don't understand how this could work, I'll copy my source and perhaps a reader can explain:
"(The yard) is hoisted to the mast-head by means of a halyard supplemented by two lateral tackles which combine the functions of lifts and braces and may therefore be termed "lift braces". The upper end of the halyard is rove through a sheave-hole some way below the mast-head and made fast to the middle of the yard. Each lift-brace functions through a purchase fitted between the mast-head and one end of the yard, a tackle consisting of two single blocks. One end of the rope is rove through the block at the end of the yard, passes to the block at the mast and returns to be made fast to the yard-arm just inner to the first block. These lift-braces, as the name implies, also control the set of the yard; by slacking away on one and hauling on the other, the yard is veered." ("The Sea-Going Mtepe and Dau of the Lamu Archipelago," James Hornell, Mariner's Mirror, January 1941)

Differences Between Mtepes and Dau al Mtepes
            Throughout the above discussion, I've treated both types of vessels together, as the features described seem mainly to have applied to both. But there were some differences between mtepes proper and dau al mtepes, some obvious and others significant but not apparent.
            The stem and stern of a dau al mtepe were raked considerably more than those of a mtepe, so that, for boats of identical length, the dau would have a far shorter keel. The stem of the mtepe was surmounted by a large curved terminal that may have represented a camel's head. The dau had no such device, but instead had a long bowsprit. The mtepe was more colorful and far more highly decorated, with occuli both fore and aft, painted bands on the topmost strakes, and amulets, tassels and flags. Perhaps by the time the dau came into being, the trade was already under competitive pressure from other, more modern boat types, and decoration was discarded as uneconomic.
Stepped, stacked breasthook stem in the dau al mtepe. The outer false stem (a) was not always present. (From Hornell)
            The most interesting difference, however, was in the construction of the stem and sternpost. The mtepe had fairly conventional timber posts, except that the stem was in two pieces joined by an unusual mortise-and-tenon joint open at the front and rear. The dau al mtepe's was even stranger: it consisted of a series of stacked, V-shaped breasthooks, each one about as tall as a strake, tapering aft, and pegged to one another. In some cases, a solid timber was fastened outboard of this stacked construction. It is not known how planks were fastened to the stems/sternposts in either case, or if the timbers were rabbeted.

A Mtepe Replica
The replica mtepe Shungwaya in the House of Wonders Museum, Zanzibar (Source: Wikipedia)
            In 2003, a mtepe was built for an exhibit on dhow culture at the House of Wonders Museum in Zanzibar. The design was to be a roughly half-size (but not half-scale) interpretation of a mtepe that had been measured in 1877 at 97 feet long, 24 feet beam, and depth of 9.5 feet (apparently including draft plus freeboard). The boatbuilder who was contracted for its construction it had never worked on a mtepe, but had heard about some of the construction processes from his grandfather, who was one of the last builders of dau al mtepes. Even though guided by academics and museum personnel, the builder diverged from several known aspects of mtepe construction, most notably in building the boat frame-first as opposed to shell-first, and in refusing to cut off the plank lashings on the hull's outer surface. Named Shungwaya, the boat was briefly tested in Zanzibar harbor, where it proved to be quick and to leak somewhat as expected. It is now on display at the House of Wonders Museum.
The mtepe Shungwaya

Main sources: 
"Construction and Qualitative Analysis of a Sewn Boat of the Western Indian Ocean, Robert Marshall Adams, (MA thesis, 1985, Texas A&M Univ.)
"Arabia to China -- the Oriental Traditions," Jeremy Green, in The Earliest Ships: The Evolution of Boats Into Ships (Conway's History of the Ship);(Naval Institute Press, 1996)
"The Sea-Going Mtepe and Dau of the Lamu Archipelago," James Hornell, Mariner's Mirror, January 1941
"The Mtepe 'Sungwaya' Sails Again," Professor Abdul Sheriff, et al, Ziff Journal, 2006


  1. What about Taratai? 75' hull, all sewn planks, built in 1976, did a lot of sea miles.
    The book is available at Amazon and the Taratai still exists today in the Auckland maritime museum.

  2. Gary - Thanks for this link. I wasn't familiar with Taratai, but she looks lovely in the video and I'll read up on her -- maybe do a post on the subject if I find enough to write about.

    When I referred to the mtepe as "the world's last sewn vessel," I meant the last one sailing as a *type* -- in regular traditional use. I didn't mean to imply that there were no replicas or rare individual survivals. There was also Tim Severin's sewn dhow Sohar, which I wrote about here:

    1. There were at least six of these "Baurua" of this size or larger in Kiribati during the 1930's, but the arrival of WWII put an end to the building and sailing.

  3. Bob,
    I can see how the sailing rig as described could be made to perform to windward efficiently. The "lift braces" would serve as downhauls in reverse. By lifting the windward end of the yard and hauling on the fore-guy (as shown, it looks like there are two), at the same time slacking the leeward lift-brace, the sail could be trimmed more along the fore and aft axis of the boat. This method would keep the decks clear of tackle run from the head of the sail to somewhere on deck where it would likely be a hazard. Without the lift-braces and fore-guys, hauling on the clew alone would not trim the top of the sail in enough for the boat to point to windward.

  4. Would the somewhat raked bowsprit (good angle for bowlines) and the apparent bowlines be key to windward performance?

  5. My ignorance of rigging, already probably obvious to some readers, is going to show up again. I'm going to assume thhatt what Wade calls the bowline is the same as what Doryman calls the fore-guy -- i.e., the lines from the leech/luff/vertical edges of the sail that lead forward to the bowsprit or stemhead. If my assumption is correct, then that line would certainly seem to be critical to the boat's pointing ability. About the angle of the bowsprit I can't say. It's interesting that on the mtepe model at the Kensington Science museum, the stemhead terminal is so high that the bowline that leads to it is horizontal, while on the second and third photos, of dau al mtepes, the bowsprit doesn't rise nearly that high.

    1. Bob,
      I want to send you an emailed article without joining Facebook. Can I do so?

    2. You may email me at (note the two "com"s, one before, one after the dot).