You are on page 1of 7

Replicating a Caravel

Author(s): Donald H. Keith


Source: Historical Archaeology, Vol. 26, No. 4, Advances in Underwater Archaeology (1992),
pp. 21-26
Published by: Society for Historical Archaeology
Stable URL: http://www.jstor.org/stable/25616190
Accessed: 10-03-2017 13:59 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Society for Historical Archaeology is collaborating with JSTOR to digitize, preserve and extend access to
Historical Archaeology

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
DONALD H. KEITH determine the rising and narrowing of the ship's
frames that will take place along the length of the
hull fore and aft of the mastercouple. These com
Replicating a Caravel plex curves, in turn, give the hull its sea-keeping
qualities, its personality.
The shipwright did not invent either the tech
ABSTRACT nique (sometimes called "Mediterranean whole
molding") or the particular proportional relation
The building of discovery-period ships for the specific pur ships incorporated in the patterns he is using. He is
pose of testing the accuracy of carefully determined but
following a set of design and construction specifi
hypothetical designs seems to be a more useful approach to
replication than fanciful celebratory inventions. Before an
cations derived from historical, ethnographical,
experimental replica ship of any type can be built, a number and archaeological sources that will result in the
of prerequisites must be met: Sufficient reliable information reproduction of a caravel. The specifications, sup
must be available. All avenues of research?archaeology, plied by caravel builder John Sarsfield (Figure 2),
history, and ethnology?must be exhausted. An inventory
call for Mediterranean whole molding to be used
of the technology available to the original shipwrights must
be taken. This article outlines the process of replicating an because historical sources imply that the technique
early 16th-century caravel from the formation of a reliable was present in Spain during this time (Sarsfield
hypothesis through the process of construction, fitting out, 1984, 1985a: 1). The expertise in how to use the
fine-tuning, and testing. technique comes from the ethnographic present,
from a master shipwright who learned it from his
father. The particular proportional relationships
Introduction contained in the scales are derived (insofar as pos
sible) from the archaeological remains of actual
On the bank of a muddy river just outside Va Spanish shipwrecks of the period.
lencia, Bahia, Brazil, a ship is being born. The time Sarsfield's Nina, really a reproduction of a ge
has come for the shipwright to define the shape of neric late 15th-century Iberian caravel, is being
the hull. He has laid the keel and erected and at built not in Spain or the United States, but in Va
tached the stempost, sternpost, dead wood, and lenca, Brazil (Figure 3). Nina is a test-bed for de
knees. On the keel's upper surface he has marked sign templates, the correct use of traditional con
the locations where the mastercouple and principal struction features and materials, and arrangements
frames will be set. Half way between the master and manipulation of sailing rig?an exercise in ex
couple and either end of the keel he has marked the perimental archaeology rather than a commemora
locations of the primary molding frames. At this tive decoration.
point the ship has no dimension other than length. Sarsfield selected Valenca for three reasons.
It can be slender, fast, and perhaps a bit tender, or First, shipwrights there commonly use Mediterra
full-bodied, stable, and stately. But it will be nei nean whole molding in conjunction with mechan
ther. The shape of the ship-to-be's hull is not de ically-generated geometric progressions known as
fined by a set of lines drawings or even by the graminhos: techniques that may be similar or iden
hand, eye, and memory of the shipwright?he has tical to those used by the builders of discovery
never made a ship like this before. No, the shape is period ships (Sarsfield 1985a:3-10, Figures 2-6).
contained in three thin wooden patterns that he will Second, Valencan shipwrights still rely on tradi
use to determine the curvature of each frame. tional tools and construction methods. The sur
The patterns are marked with parallel lines vival of such anachronistic practices provides an
spaced according to a geometric progression de under-utilized but important avenue of research.
rived mechanically from arcs of a circle (Figure 1). Finally, the tropical forests of Bahia provide a
The only tools needed to make these patterns are a source for the various naturally-shaped timbers
compass and a straight-edge. Together the patterns necessary to build large wooden ships.

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
22 HISTORICAL ARCHAEOLOGY, VOLUME 26

FIGURE 1. Graminhos scales used by John Sarsfield to


determine the shape of Nina's hull using the technique of
Mediterranean whole molding. (Photo by Toni L. Carrell.)

Combining ethnographic survivals with perti


nent results of documentary, artistic, and archae
ological research, Sarsfield's Nina will likely be
the most authentic reproduction of a Columbus-era
ship ever built. As such, it provides maritime ar
chaeologists and historians with an opportunity to
take advantage of the fifth avenue of research, ex
perimental archaeology. As Carrell (this volume) FIGURE 2. John Patrick Sarsfield (/e/r), caravel builder.
has pointed out, reproducing a ship for the purpose (Photo by Toni L. Carrell.)
of better understanding how it was made, why it
was built the way it was, and for testing its per
formance at sea is a relatively new phenomenon,
one that few archaeologists have explored. But in level of significance to attach to the myriad of tool
recent decades reproductions such as the trireme marks, nail holes, scribe lines, bevels, recesses,
Olympias and the Kyrenia ship have been built for and other consequences of construction that, when
the specific purpose of testing the accuracy of care they appear on excavated shipwreck remains, beg
fully determined, but hypothetical, reconstruc explanation.
tions.
But what distinguishes an exercise in experi
mental archaeology from merely building a rep
lica? The most important difference lies in the pur
Replication and Experimental Archaeology pose (Figure 4). An exercise in experimental
archaeology begins with establishing the purpose
In addition to the primary benefit of experimen of the experiment. The parameters of the test must
tal archaeology, namely using the completed re be stated. Given the purpose of the experiment and
production to test hypotheses, other enlightenment the parameters of the test, the corpus of available
results during the process of replication. The ex data must be evaluated to determine if sufficient
perimental archaeologist is likely to acquire a re information to accomplish the purpose of the ex
fined appreciation for the technical minutiae of de periment is available. If the answer is no, the ex
signing and constructing a ship, including what periment or the parameters of the test must be

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
REPLICATING A CARAVEL 23

15th-century caravel for the purpose of determin


ing its seaworthiness, it must first be established
that the test will require a certain number of voy
ages across the Atlantic Ocean during a variety of
seasons and sea states. Provision will be made to
test a number of variables such as mast locations,
rigging, sail configurations, and hull trim in a con
trolled manner. Performance will be evaluated for
ability to sail to windward, stability, speed, capac
ity, and durability. Other information can be gath
ered on maintenance requirements, optimum crew
size, and provisioning.
FIGURE 3. A//71a under construction in Valenga, Bahia, Is there sufficient knowledge about discovery
Brazil in June 1990. (Photo by Toni L. Carrell.) period ships in general and caravels in particular to
enable the construction of a hypothetical reproduc
HYPOTHETICAL REPRODUCTION BY EXPERIMENT tion? Have all avenues of research been exhausted?
What were the ship's gross dimensions, the shape
Define purpose of experiment of the hull, how that shape was arrived at by the
Define parameters of test
shipwright, the internal and external structural ar
Determine sufficiency of information
Define limitations of use of modern technology rangements, the techniques and materials used to
Build reproduction build it? Is there an inventory of the technologies
Test reproduction available to the original shipwrights? Obviously,
Evaluate results
there are gaps in present knowledge?if everything
Refine experiment
Repeat already was known about the ships there would be
no reason to experiment?but the gaps must not lie
FIGURE 4. Steps toward the creation of a hypothetical in critical areas. Assessing the state of current
reproduction by experiment.
knowledge, there is really quite a lot known about
exploratory-period ships in general?but a good
deal less about caravels in particular.
From written history?tax documents, invento
modified. If the answer is yes, determinations of ries, legal papers, and the like?are obtained rel
the extent to which modern technology and knowl ative sizes, proportions and ratios, number of
edge can be used without compromising the exper masts, materials used in construction and their
iment must be made. At this point, work on the cost, length of time required to build a ship, size of
hypothetical reproduction can begin. crew required to sail a ship, numbers and types of
Once the ship is built, the experiment can be objects considered to be part of a ship's equip
conducted. At this point the exercise moves into ment, as well as where ships were built and by
another realm altogether?actual voyaging?dur whom (Lyon 1986:600-605; Lakey 1989:88-90).
ing which the capabilities and limits of a hypothet From art history?models, paintings, reliefs,
ical ship reproduction will make themselves appar tapestries, graffiti, sketches, and drawings?one
ent immediately, and it will be up to the can see that these ships appear to possess features
experimental archaeologist whether to assign the such as a small but capacious hull; at least three
various apparent assets and liabilities to the origi masts and a bowsprit; a combination of square
nal, or to flaws in the reproduction process. Sub rigged and lateen, or triangular, sails; low castles
sequent reproductions can build on the successes fore and aft; a sternpost rudder; and heavy timbers
and failures of previous experiments. running horizontally and vertically outside the hull
If, for instance, one intends to reproduce a late (Clowes 1927; Keith 1989:87).

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
24 HISTORICAL ARCHAEOLOGY, VOLUME 26

From ethnography derive fundamental wood


working technologies, anachronistic methods of
hull design, wood selection folklore, and some in
sight into the tenuous connection between material
culture and behavior (Sarsfield 1985b:87-89).
From archaeology come interior structure (par
ticularly of the bottom of the ship), iron and bronze
^^^^^^^^
rigging components, fastener types and patterns,
materials used in the fabrication of the hull
(woods, metals, etc.), absolute dimensions, and
actual examples of items of a ship's equipment,
stores, and cargo (Bradley 1981; Keith 1989:90
92;Oertling 1989:100-103).
What is missing is complete, reliable (preferably
archaeological) information that would allow the FIGURE 5. Trapezoidal joint between floors and futtocks
used on Nina. (Drawing by T. J. Oertling.)
replication of hull shape; anything at all about the
interior partitioning and accommodations for crew
and officers; and of course any idea whatsoever of
the seaworthiness, performance, and handling
characteristics of caravels. of certain frames in archaeological examples of
Is this sufficient information? Probably, with the discovery-period hulls (Figure 5), Sarsfield was
important acknowledgment that the information in called away from Valenca for a few days. When he
one critical area, true hull shape, is deficient. In returned, he discovered that his shipwrights not
any case it is the best possible until a bona fide, only mastered this complicated joint, they liked it
well-preserved caravel from the discovery period so much that they used it between every floor and
is found, excavated, analyzed, and reconstructed. first futtock throughout the ship! They also began
Complete, authentic replication implies that the to talk of using it in the construction of other com
experimenter must know and go through all the mercial hulls they plan to build in the future. If a
steps used by the original builders. As it is patently revival of this ancient technique results, it will
impossible to replicate everything?design proce constitute a curious case of an extinct technology
dures, tools, materials, techniques, craftsmanship, cross-fertilizing the present.
decoration, environmental ambiance, attitudes, Different types of materials have vastly different
and number of people involved?the experimental properties?density, flexibility, durability, work
archaeologist will have to be selective. This leads ability. This is especially true of wood. While it is
to judgment calls regarding what is important to clearly anathema for the experimental archaeolo
replicate and what will not make any difference gist to consider making an authentic caravel's hull
and therefore can be sacrificed. out of ferro-cement, what does one do when tim
For example, is it necessary for the purpose of bers long enough, thick enough, of the right spe
the experiment to use ancient technologies cies of wood, and in sufficient quantities simply no
throughout the building, or is it enough to test a longer exist on the face of the earth? The oak keel
variety of techniques just long enough to satisfy of the Viking ship replica built in 1893 was im
curiosity before reverting to shortcuts made possi ported from Canada because no tree of sufficient
ble through the use of modern technology? An size could be located in Norway (Sjovold 1985:
interesting reversal of this process occurred during 58).
the building of Sarsfield's Nina. After showing his Decisions such as these must be made on a case
shipwrights how to replicate the curious trapezoi by-case basis. Generally speaking close adherence
dal joint found between the floors and first futtocks to authentic detail is ultimately the safest course of

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
REPLICATING A CARAVEL 25

action, leaving the experiment free of the confu step in any comprehensive reconstruction of the
sion that can result when several variables are past.
changed simultaneously. Pekka Toivanen (1991,
pers. comm.) has observed that even apparently
insignificant digressions from traditional proce Discussion
dures can have profound effects: timbers shaped by
adzing and axing are more resistant to rot than Exploratory-period ships in general, and carav
timbers shaped by sawing, pitching a hull below els in particular, have probably been reproduced
the waterline can add a knot or two to a vessel's more often than any other types of ships. Unfor
top speed by reducing drag, and authentic pine tar tunately, most of these ships have been created for
is faster than coal tar. celebratory purposes, and consequently little has
Can synthetic materials, plastics, metal alloys, been learned from the difficult and costly pro
and composites actually do a better job of replica cesses of building and sailing them. A century ago,
tion than the use of original materials, or will the when the world celebrated the 400th anniversary of
purpose of the experiment be compromised? Is it Columbus' voyage, recreations of Santa Maria,
absolutely essential that through-the-hull fasteners Pinta, and Nina were built in Spain. A measure of
below the waterline be gasketed with twists of how inaccurate those recreations were can be
tarred hemp, or will other fibers do as well? Ar gleaned from the fact that the two caravels, Nina
chaeological evidence demonstrates that explor and Pinta, had to be towed across the Atlantic.
atory-period hulls were fastened exclusively with Santa Maria sailed across under the close and
wrought-iron bolts, pins, and nails. Today, it is skeptical escort of steamship chaperons, but it was
prohibitively expensive, if not impossible to obtain slow and dangerously unstable. Yet, the tens of
wrought-iron fasteners and fittings in sufficient thousands of people who toured these ships left
quantity to build a caravel. Consequently, the with the impression they had seen authentic repli
Sarsfield reproduction relies on made-to-order gal cas.
vanized mild steel fasteners. To what extent will Almost without exception, today's recreation
substitutions such as this compromise the validity (and there are at least eight) are imitations of tho
of the experiment? Again, it all depends on the built for the 1892 celebrations?replicas of rec
purpose of the experiment, but the point is moot if
ations! Concessions to safety, such as the insta
there is no better alternative. tion of diesel engines, satellite navigation, and w
Despite all this attention to authenticity, ter-tight bulkheads, are of paramount importan
problems arise. All too frequently the data of to most recreations due to their extremely limit
history, archaeology, and ethnography appear to capabilities. During sea trials last year the "o
contradict each other or are mute on an important ficial" Columbus fleet was able to achieve a ma
detail. Was the interior divided into compart imum headway of only four knots, even when u
ments? Was the standing rigging attached inside ing both sails and engines. The solution? At la
the gunwale, as it is in the most detailed report they were being re-fitted with more powe
three-dimensional representation of the period, the ful engines!
Mataro model, or outside the hull, in the more John Sarsfield viewed the taking of design and
modern convention? What was the configuration construction shortcuts and the addition of modern
of the superstructure? What accommodations were amenities as tacit admission that the builders them
made for the removal of human waste? What selves do not really believe that the ships they
kinds of ordnance did it carry, and where were the claim to be replicating were ever safe or seawor
different types of guns mounted? Although the thy. In contrast, he believed that discovery-period
answers to some of these questions can be worked caravels were as seaworthy and reliable as any ship
out from models, the majority can be obtained that ever sailed the seas. He was right. His Nina
only through experimental archaeology?the next outperformed the competition in the "1992 Co

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms
26 HISTORICAL ARCHAEOLOGY, VOLUME 26

lumbus regatta," setting an example for hypothet Lyon, Eugene


1986 15th-century Manuscript Yields First Look at Nina.
ical reproduction projects and pointing the way to
National Geographic 170(5):600-605.
a new standard, one that should eventually sup
plant fanciful celebratory recreations altogether.
Oertling, Thomas J.
1989 The Few Remaining Clues. In Underwater Archae
ology Proceedings from the Society for Historical
REFERENCES Archaeology Conference, edited by J. Barto Arnold
III, pp. 100-104. Society for Historical Archaeol
Bradley, Charles ogy, California, Pennsylvania.
1981 Ship's Fittings and Rigging Components from Un
derwater Archaeological Excavations at Red Bay, Sarsfield, John P.
Labrador. The Red Bay Project: Interim Report, 1984 Mediterranean Whole Moulding. Mariners Mirror
1981. Parks Canada, Ottawa. 70(l):86-88.
1985a Survival of Pre-Sixteenth Century Mediterranean
Clowes, G. J. Lofting Techniques in Bahia, Brasil. Paper presented
1927 Ships of the Early Explorers. Mariners Mirror 12(1): at the Fourth Meeting of the International Sympo
216-235.
sium on Boat and Ship Archaeology, Lisbon.
Keith, Donald H. 1985b From the Brink of Extinction. Wooden Boat 66:84
1989 The Mysterious Caravel. In Underwater Archaeol 89.
ogy Proceedings from the Society for Historical Ar
Sjovold, Thorleif
chaeology Conference, edited by J. Barto Arnold III,
1985 The Viking Ships in Oslo. Universitetets Oldsaksam
pp. 90-92. Society for Historical Archaeology, Cal
ling, Oslo.
ifornia, Pennsylvania.

Lakey, Denise C.
1989 Historical Archaeology. In Underwater Archaeology
Proceedings from the Society for Historical Archae Donald H. Keith
ology Conference, edited by J. Barto Arnold III, pp. Ships of Discovery
88-90. Society for Historical Archaeology, Califor P.O. Box 542865
nia, Pennsylvania. Dallas, Texas 75354

This content downloaded from 193.198.212.4 on Fri, 10 Mar 2017 13:59:31 UTC
All use subject to http://about.jstor.org/terms

You might also like