The Outriggers of Indonesian Canoes. Author(s): A. C. Haddon Source: The Journal of the Royal Anthropological Institute of Great Britain and Ireland, Vol. 50 (Jan. - Jun., 1920), pp. 69-134 Published by: Royal Anthropological Institute of Great Britain and Ireland Stable URL: http://www.jstor.org/stable/2843375 . Accessed: 24/06/2014 21:30

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

. 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 [email protected].

.

Royal Anthropological Institute of Great Britain and Ireland is collaborating with JSTOR to digitize, preserve and extend access to The Journal of the Royal Anthropological Institute of Great Britain and Ireland.

http://www.jstor.org

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 69

THE OUTRIGGERS OF INDONESIAN CANOES.

By A. C. HADDON.

CONTENTS. PAGE Material ...... 70 Terminology ...... 72 Double Canoes ...... 77 The Distribution of Single and Double Outriggers ..., ...... 78 The Number of the Outrigger-Booms ...... 79 The Attachment of the-Booms to the Hull ...... 82 The Float ...... 83 The Attachmentsbetween the Booms and the Float and theirDistribution A.-Direct:- 1. Inserted ...... 83, 123 2. Lashed ...... 83, 124 3. Mixed Direct Attachment ...... 84 B.-Indirect: (a) Attachmentinserted into the Float 1. Stick ...... 84, 125 2. Ro d ...... 87 3. Spike ...... 79, 87, 127 4. Y-shaped Stick ...... 87, 127 5. Board ...... 87, 128 6. Balinese ...... 73, 88, 128 (b) Attachmenttied to the Float:- 1. Moluccan ...... 89, 129 2. Halmaheran ...... 90, 129 C.-Mixed Attachments:- 1. Mixed Direct and Balinese ...... 84, 92 2. Mixed Direct and Rod ...... 92 3. Mixed Direct and Halmaheran ...... 94 4. Mixed Rod and Halmaheran ...... 96 Notes on the Characteristic OutriggerCanoes of the Main Districts of Indonesia ... 97 Notes on the Native Names for Canoes and Outriggers ...... 116 General Conclusions ...... 119 Bibliography ...... 131

For several years I have been accumulatingmaterial dealing with the outrigger canoes of Indonesia, of whichI now presenta brief and generalaccount, though I am fullyaware of the imperfectionof my data. Much more remainsto be done by observationsin the field,by a more exhaustivetreatment of the literatureon the subject,and by a study of photographsin the librariesof certainlearned societies

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 70 A. C. HADDON.-The Outriggersof Indonesian Canoes. and museums,as wellas of specimenspreserved in numerousmuseums in variousparts of theworld. I hope,however, that thisessay will help towards a moresystematic con- siderationof the details of constructionand of the problemsof culturaldistribution.

M,ATERIAL. A wordof cautionis necessarywith regard to theavailable data. A photograph, or an illustrationmade froma photograph,is good evidencethat a certainform was then presentin a given spot, but it is not conclusivethat it is the usual or common formthere, as any canoe may be a chance visitor. It is only when a photograph shows numerousexamples of the same type that it becomes authoritative,or when differentphotographs agree in this respect. Of far greatervalue is the explicit statementof a traveller,and I have endeavouredin all cases to recordsuch statements. In drawingsor sketchesmade by travellersa greatdeal dependson the skill of the drauglhtsman,his keennessof eyesight,and his knowledgeof the details of con- struction. The unsatisfactorycharacter of illustrations holds good formodels, except that the commontype is far morelikely to be represented. But a furthersoufrce of errorawaits one. The model may be carelesslymade and certaindetails may be slurredover or even wronglyconstructed to save trouble. On the otherhand, the models as a rule seem to be made by those thoroughlyconversant with the vessels, and probably,in many cases, are made by those who actually build boats. The technical skill in model-makingwhich is characteristicof these peoples, and the pride of the artificer,however, give one confidencein acceptingmodels as trust- worthy,even though the variousparts may not be made exact as regardstheir relative proportions. Bearingin mind these limitations,I have not shrunkfrom accepting modelsas good evidence. In a considerablenumber of cases the writtendescriptions by travellersare far fromsatisfactory, as they oftendo not appreciatethose details whichare of interestto a specialist; frequentlyno descriptionat all is vouchsafed, and we have thento relysolely on any illustrationthat may be supplied. Several of the followingmuseums are referredto by the names of the towns in whichthey are situated: Amsterdam,Konitnklijk Zoologisch Genootschap. EthnographicalMuseum of the Natura ArtisMagistra. BritishMuseum, London. Gambridge,Museum of Archweologyand Ethnology. Edinburgh,Museum of Science and Art. Halifax, BankfieldMuseum. HornimanMuseum, Forest Hill, London,S.E. Leiden, Rijks EthnographischMuseum. Oxford,Pitt-Rivers Museum. Rotterdam,Ethnologisch en MaritiemMuseum. Salem (Mass., U.S.A.), Peabody Museum.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions +++tt Limit of sinZe- ouatre,e-s \oo oPapo -iqistra/c?zn ared ofdoitce oatrijirer$wim -/ e' g; /tro1ors. 54tcAk hnetatscoitic&<' .Z t _ MantDAeo++, 4d MaftL ~~~~~~~wil~kAeayas wet4sed1y?aL g + e1 } 9 y z t @p% cxcladay Nti.s. ut i- Are of6oZ yersritmo rLt W9 u

+ \\ 9\ 8 ...... / ...... ; '\ Direct laske-d afta4uneartt fD' S,,,0-TQLUR _,.h Lae-rcvt-

C,~~~~~~~~~~~~~~~~~~C

lIE NTAVY&E 1-~~+oR +g ~tSAfl ~ d7 C 0BAWE4AN 5BUTON ~ j\KS

DJSTRIDUTIONOF INDONESIAN OUTRIGCERS

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 72 A. C. HADDON.-The Outriggersof Indonesian Canoes.

I take this opportunityof thankingvarious friendsand colleagues who have helped me with information,and would more particularlylike to mentionHenry Balfour,Ivor H. Evans, Dr. G. Friederici,Dr. F. H. H. Guillemard,Lawrence W. Jenkins,Dr. H. H. Juynboll,H. Ling Roth. For convenienceof referenceI shall, as a general rule, mentionthe several islands and places in the followingorder: Andamans. Nicobars. Nias, Mentawei Islands, and Engano (islands to the west of Sumatra). Sumatra: Palembang (in the south-eastof Sumatra). Java. Madura. Bawean Islands (northof Madura). The Lesser Sunda Islands (Bali to Tenimber): Bali; Lombok; Sumbawa; Sumba (Sandalwood Island); Savu; Flores; Solor; Lomblen; Ombai (Allor); Timor; Rotti; Wetta (Wetteror Eetar); Serwatti; Baba (Babber); Tenimber(Tanimbar) or TimorLaut (Timorlao)Islands. Kei (Ke) Islands. Aru Islands. The Moluccas (Banda to Halmahera): Banda; Buru; Amblau; Amboina (Ambon); The Uliasser Islands: Haruku, Saparua and Nusa Laut; Ceram (Seran); Ceram Laut; Goram; Misol; Obi (Ombi); Batjan; Tidor; Ternate; Halmahera (Gilolo): Weda Bay (the southernGulf of Halmahera), Ake Selaka, Tobelo, and Galela (in the northof Halmahera). Xulla (Sulla) Isla4ds. Butong (Butung). Salaier. Celebes: Makassar and Gowa (Goa), Gulf of Boni (the southernGulf of Celebes),Tontoli (Toli-Toli,north-west Celebes), Minabassa (the north-eastpoint of Celebes) with the towns of Kema and Menado, and the island of Limbe. Banka. Talisse. Talaut Islands (includingthe large island of Sangir). Tulur (Tulaur or Salibabu) Islands. Nanusa Islands. Sulu Archipelago. Philippines: Mindanao (Zamboanga at its south-west point), the Gagayanes (Cagayan) group,Zebu (Cebu), and Manila. Borneo: Sarawak, a raj on the northcoast. North-westarea of New Guinea: Skroe on the south and Sekar on the northside of the Onin Peninsula; Waigiu (Waigeu); Saonek (an islet offthe south shore of Waigiu); Sorong (a village just south of the extremenorth-west point of New Guinea). GeelvinkBay. (See Map, p. 71.)

TERMINOLOGY.

To avoid ambiguityI proposeto adopt the followingterms as here defined:- The outriggeris a balancingapparatus that extendstransversely across the hull of the canoe; the transversepoles of an outriggerare outrigger-booms(or simply booms),their free extremities may be attacheddirectly to thefloat,1 or indirectlyby variousmethods; in all cases thisis spokenof as the attachment.Various methods of indirectattachment will be sufficientlydescribed in the course of this paper. They consistin the main of bent ratan in the Moluccanattachment, or of a stickor sticks, a rod, or of a variouslyshaped piece of wood, the outrigger-sparor attachment-spar

1 Frequentlyauthors speak of the floatas an " outrigger."

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 73

(or simplyspar). A thin spar, bracingspar, may pass fromthe attachment-spar boom; this may be called a complex attachment. In many outriggersthere are one or more poles connectingthe booms and usually lashed above them, but sometimesbeneath them. These may be termed .longitudinalspars, as theyare longitudinalto, that is, parallelwith, the lengthof the canoe,though transverse as regardsthe lengthof the booms themselves. They might thereforewith justice be termed" transversespars," but as the booms themselvesare transverseas regardto thecanoe and theyare at rightangles to the lie of the booms, it seemspreferable to adhereto myformer term. Sometimesthere is a spar running more or less midway betweenthe hull and the attachments-thismay be termed -thecentral longitudinal spar; the outerand innerlongitudinal spars run respectively immediatelyon the outerand inneraspects of the attachment(Figs. 16 B, 17). In Micronesiaa spar may pass diagonallyfrom the hull to the end of.the outrigger apparatus-this I terma stayspar. Boom-prolongation.-Thissomewhat cumbersome term may be applied to a type of boom that appears to be confinedto the East Java-Lombok area. The boom properis shortand thick,but a thinnerspar is attachedto it, the freeend of which is eitherinserted into or lashed to the floator connectedwith it by means of a rod- attachment. In small models of canoes fromthis area the prolongationis inserted into the boom, but the exact method of the junction in actual vessels has been ,describedonly in the case of the sedek,and the publishedillustrations known to me are not clear on this point. It may be suggestedthat the boom was made thickin orderto strengthenthe hull,but in that case it would obviouslybe too heavyto form an efficientoutrigger; to remedythis the projectingportions might be thinneddown (whichpossibly may occur)or a moreslender spar of wood or bamboo mightbe fixed on to it. The sedekof the Balinese attachment(p. 88) is thus merelyone formof a boom-prolongation. Occasionallythere is a central outrigger-boomwhich is lashed to the various longitudinalspars but is not connectedwith the floatby means of a regularattach- ment,though it may be tied to it by means of a long lashing-this may be termeda false-boom. Occasionally,outside our area, morethan one false-boommay occur. Theremay be above a boom anotherspar whichvaries greatlyin length,some- timesbeing as long as the boom itself,in whichcase it mightbe regardedas a boom, thus givingrise to an accessoryor double boom parallel to the boom proper; but in orderto make the termapplicable to all its variations,I preferto describeit as a boom-spar. A boom-spar,one end ofwhich rests on theroof of a shelterof a Mentawei

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 74 A. C. HADDON.-The Outriggersof IndonesianCanoes.

war-vessel,knabat bogolu (Fig. 1), is figuredby Rosenberg(1888, P1. XVIII, Fig. 9). Baessler (P1. VIII, Fig. 4) gives a drawingof a model of a sailingcanoe fromWetta

FIG. 1.-MODEL OF A WAR VESSEL, knabatbogolu, MENTAWEI ISLANDS (AFTER ROSENBERG, 1888); THE RIGGING OF THE VESSEL IS OMITTED. x

C

D

FIG. 2,-SECTIONS OF MODELS OF CANOES: A, SULU; B, MORO CANOE, ZAMBOANGA; C, MORO CANOE, ZEBU; D, SANGIR (FROM MULLER, 1912, figs.55, 56, 57, 58).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 75

(Wetter) Islands with a double outriggerof two straightbooms, above which are equally long straightboom-spars, these evidentlycorrespond with those just men- tioned; but in this case thereis a U -Moluccanattachment, the upperlimbs of which are lashed to its boom and boom-spar(Fig. 3). In a model canoe fromTenimber in

FIG. 3.-MODEL OF A SAILING SHIP, WITH TRIPOD MAST, STRAIGHT BOOMS AND BOOM SPARS AND A U-MOLUCCAN ATTACHMENT, WETTA (FROM BAESSLER).

EIG. 4.-MODEL WITH A BOOM SPAR AND HALMAHERAN ATTACHMENT, TENIMBER (LEIDEN).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 76 A. C. HADDON.-The Outriggersof Indonesian Canoes. the Leiden Museumis a thinboom-spar above each true boom; it lies over a longi- tudinalspar whichrests on the two boomsand its end is lashed to the upperfork of a Halmaheranattachment (Fig. 4). Muller(1912, p. 244) describesa modelof a canoe fromSangir in whichthe two outer (foreand aft) booms and theirboom-spars are attached directlyto the float,presumably by a lashing; the innerbooms have the arrangementshown in Fig. 2, D, above each of these booms is an equally long and slenderboom-spar which turns up at its ends; attached to the ends of the boom and boom-sparis the long stem of a l-shaped attachment-spar,beneath which the doublefloat is fastened; the boom-spar,boom, and attachment-sparare strengthened by a bracingspar; obviouslyall these,elements are lashed together. At Kema, NorthCelebes (G., No. 232), the Sulu Islands (Guillemard,1889, p. 192; G., Nos. 4, 115; Muller,Fig. 55; Savage Landor, II, pp. 2, 12), Zamboangaand Zebu in the Philippines(Muiller, 1912, Figs. 56, 57), the upwardlycurved boom-sparis greatly reducedin lengthand may becomebowed and have carvedends. It does not support the float,but is attached by lashingsto the unusually thin and fragileoutrigger- boom which is thus strengthenedby the elasticityof the bowed boom-spar. As previouslymentioned, the outrigger-booms,wbich usually are moie than two in number,are attacheddirectly lashed to the float. Muller(p. 245) describesthe trans- formationof supernumerary boom-spars into curved crescentic ornaments in theSangir and Sulu Islands; but these can no longerbe termedboom-spars, as they are not necessarilyconnected with outrigger-booms. Outlayers.-In describingthe " Flying Praws " of Borneo,D. Beeckman says: "To preventtheir oversetting, they fixtwo long Poles or Outlayersone across the Fore-partand anotherat the After-partof the Boat; each end beingrun into a large Bamboe . . . and whenit blows hard, the People run out and in on the Outlayers, accordingas the Gale is fresheror abates, to keep the Boat upright" (1718, p. 91). Evidentlythis is what is referredto by R. Munday (1848, I, p. 52), quotingfrom [Raja] J. Brooke'sJournal of January,1840: " The small sailingboats [at Palette, Gulfof Boni, Celebes] had outriggersof wood, which weighted with men, enabled them to carrya sail of enormoussize." Folkard says that the canoes at Manila have no outriggers," but merelyan outlager,or pole, laid across the vessel amidships,and extendingseveral feet beyond the sides " (f.n., p. 482). Lane Fox (Pitt-Rivers) points out (1875, p. 430) the practicalutility of a singleoutlayer for canoes which have but a singleoutrigger; he termsit a weatherplatform, when a flooringhas been laid across the booms. He adds: " We have, in the Asiatic Archipelago,a con- trivance which may be said to be derived partly fromthe double outrigger,and partlyfrom the weatherplatform. . . . A weatherplatform had beenfound sufficient to balance the vessel on one side, and the next step was to knock offthe outrigger log [float]on the otherside, thereby converting the outriggerplatform into a weather platform,the two platformsprojecting one on each side of the vessel,on the level of the gunwales,without touching the water. . . . These doubleweather-platform boats

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 77 were also found more convenientin inland waters,in the canals in Manila, and elsewhere." He also quotes the accounts of outlayersin the Philippinesby De Guines (1796) and at Manila by Dampier (1686). " Balance poles " are referredto in the Census of thePhilippine Islands, I, p. 326, Washington,1905. The outlayer may verywell be evolvedfrom an outrigger,but as in the Philippinesand Indonesia generally,the canoes have a double outrigger(when they have any at all), there does not seem to be any necessityfor the intermediatestage of a single outrigger

FIG. 5.-MODEL OF CANOE WITH A DIRECT LASHED ATTACHMENT AND AN OUTLAYER, SANGIR, TALAUT ISLANDS (AMSTERDAM). postulatedby Pitt-Riversfor this area. In a carelessly-mademodel or sketch an outlayermight very well be mistakenfor an outrigger. An outlayermay therefore be definedas a pole or a simpleframework balancing apparatus, and may be single (Fig. 5) or double. When a platformof boards or closely-laidbamboos or pol\esis formed,the outlayermay be termeda weatherplatform, which similarlymay be singleor double.

DOUBLE CANOES.

The onlyinstance known to me of an Indonesiandouble canoe is that illustrated by Hickson(1889, Fig. 10, p. 164): thisis a smallmodel used forceremonial purposes in the Nanusa Islands to the north-eastof Celebes beyond the Salibabu Islands. This sakitcanoe is composedof two canoes close together; both have equally elon- gated upturnedends and are crossedby threebooms, the ends of whichare pegged on to the floats. Mullersuggests that this idea of a double canoe may be due to Micronesianinfluence, and queries whetherthe outriggermay not universallybe traceableto a double canoe (1912, p. 239). Double canoes,janjar, are employedon the riversof the Malabar coast, South- westIndia, forconveying cattle and bulkygoods acrossthe rivers; fromthe account by Edye these appear to be temporaryarrangements (Edye, p. 6, P1. III). In the OxfordMuseum are threemodels of double canoes; one is from" India," another fromMirzapore, on the Ganges,and the thirdfrom Ceylon. The importanceof the doublecanoe in the earlynavigation of the Pacificis wellknown. That double canoes once occurredin Indonesia is extremelyprobable, and we may assumethat the double canoes of Oceania had theirorigin in Indonesia.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 78 A. C. HADDON.-The Outriggersof Indonesian Canoes.

THE DISTRIBUTION OF SINGLE AND DOUBLE OUTRIGGERS. Whereveroutriggers occur in Indonesia they are double, with the exceptions noted below (see Map, p. 71):- Single outriggersalone are found in the Andamans (Fig. 11, A) and in the Nicobars (Fig. 10, B), but thereare othercanoes in these islands whichare without an outrigger. The Sumatransailing craft known as jellorehas sometimesonly one outrigger,which is then alternatelyto windwardand leeward (Folkard, p. 481). H. von Rosenberg(1888, P1. XVIII, Fig. 19) figuresa model of a warship,knabat bogolu,from the MentaweiIslands (west of Sumatra) which appears to have but a single outriggeron the starboardside; it has two main booms whichslope down to the floatand each has an accessoryboom, whichas it slopes down fromthe roofof the cabin, can hardlybe termeda proper boom-spar; thereis a double float and apparentlya directattachment (Fig. 1). Folkard states (p. 485) that on the north coast ofJava, at Maduxa,they have sailingboats withsingle outriggers on the leeward side, while on a kind of rack on the windwardside they sometimesplace a canoe and everythingon board that is movable. An aberranttype fromthe northcoast ofmiddle Java has been notedby Hornell. Pitt-Rivers(Lane Fox) mentionsa single outriggerin Borneo,but does not give anyreference (1875, p. 429) ; fortoy canoes with a single outriggerfrom the Sarawak coast see Fig. 32. There is a model in the AmsterdamMuseum of a canoe fromSangir with a single outrigger(Fig. 5), the two booms of whichproject on the otherside of the hull to forman " outlayer"; the downwardlycurved ends of the booms are lashed to the float by a direct attachment. Numerouscanoes withsingle outriggers from various parts of Indonesia are figuredby Nicolas,De Bry and by Valentijn,but as the illustrationsgiven by the old authorsdo not appear to be always accurate,this evidencecannot be fullyrelied on; as nearly,all the boats are drawn in side-view,only one outriggercould be depicted. De Bry gives only one illustrationof a double outrigger,a coracora (V, P1. c A), whichis also figuredby Nicolas (II, P1. 9), who calls it coracora or carcolle of Banda, but Nicolas figures(P1. 14) a small war-vessel,carcolle, with one, and describes(p. 19) the carcolleof the King of Ternate as having a double outrigger. We are probablyjustified in assumingthat tbe outriggerwas doublein all theMoluccan craftin whichone was present. To the west of Indonesia canoes with a single outriggerand two booms are foundin Ceylonand parts of SouthernIndia; as earlyas 1599 Lintscotusfigured the type occurringat Goa and Cochin (II, Tab. XIV). They also extended to the Maldives. Canoes with double outriggersand two booms occur on the coast of Eastern Africa,the Comoro Islands, and North-westernMadagascar. In Madagascar the largercanoes appear to have but one float,though the boomsproject beyond the other side of the hull to forman outlayer(Haddon, 1918, No. 29). I regretthat in this paperI omittedthe record by Mullerin Madagascarof small, double-outrigger coasting

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 79 canoes, and of a large Sakalava sailingboat, lakkafiara,with a singleoutrigger, the floatof whichhas superfluous,overloaded terminal enlargements that resemblethe ends of the Hawaiian double canoe, and, Muller adds, perhaps is evidence that Malays voyagedto Madagascarin double canoes (1912,f.n., p. 239). I have recently comeacross a book by L. (1598) who gives a plate, on p. 6 (drawnin 1595),of two dug- outswith a double outriggerof two booms and a verticalstick attachment; the float appears to be a long thin log. This is the earliestrecord known to me of this kind of craftin Madagascar, and it adds to the interestto findthat it occurredin St. AugustinBay on the south-westpart of the island. To the east of Indonesia canoes with a single outriggerare the commontype in Oceania, and in New Guinea they begin to appear in GeelvinkBay and continue downthe coast (p. 122, and Haddon, 1913). The doubleoutrigger is foundthroughout GeelvinkBay and ceases at Cape D'Urville, at its easternentrance, but withinthis area the singleoutrigger predominates and, like the double-outrigger,type, has three or fourto a dozen booms. Both kinds of canoes have a new type of attachment whichconsists of a nail-likewooden spike or stick whichpasses fromabove through the end of the boom and is driven verticallyinto the float. Spikes are usually selected which have a natural thin branch projectingat right angles; this is laid overthe boom and tied firmlyto it (Fr., II, pp. 249, 252-254). Both kindsof spikes (i.e., with or withoutthe branch)appear to be used indiscriminatelyand separately, at all events at Ansus (Guillemard,pp. 401, 402, 404; G., Nos. 271, 296, 302, 439, 442). This may be termeda " spike attachment." An inner longitudinalspar is frequentlypresent (p. 127).

THE NUMBER OF THE OUTRIGGER BOOMS. South of a line whichpasses northof Borneo, Celebes, Halmahera,and Ceram the outriggersalmost invariably have but two booms,whereas to the northand east of it they have usually four,rarely more, sometimesthree and occasionallyonly two booms (Map, p. 71). In the Andamans the smallercanoes are fittedwith a singleoutrigger consisting of threeto seven booms (A. R. BrownMS.). Mouat figuresfour (p. 315), as does Folkard (p. 460). Man (XII, P1. VII) gives a photographof a small " dug-outcalled ch6-rigma" (p. 116),which has three. Modelsin the British,Cambridge, Edinburgh, Halifax,Horniman, and OxfordMuseums have three. As the early writersdid not mentionan outrigger,Mouat thoughtthat it was a recentintroduction, but Man (XI, p. 272) deniesthis. Mouat also supposedthat it was adoptedfrom the Ceylonese craft,but the type of attachmententirely negatives this wild suggestion. As we shall see, the stick attachmentis, however,very close to that of the Nicobarese canoes. There are onlytwo booms in the Nicobaresecanoes. Judgingfrom photographs (G., No. 267) fourbooms may occurat Misoland Weda

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 80 A. C. HADDON.-The Outriggersof Indonesian Canoes.

Bay (G., No. 292,Fig. 6), but usuallyin Weda Bay, Patani, and Bali, all in Halmahera, the outriggerhas two booms witha Halmaheranattachment, but frequentlythere is a central"false boom," the end of whichis connectedonly by a lashingto the centre

G =_Q<~~7,-

FIG. 6.-CANOE WITH DOUBLE OUTRIGGER, FOUR BOOMS, AND HALMAHERAN ATTACHMENT, WEDA BAY, HALMAHERA (Photo G., No. 292).

of the float (Friederici,II, p. 242). Forrest figuresa " Molucca Corocoro" (pL. 4, p. 82) withthree booms offBatjan (cf.p. 110), and a similarvessel at the Kanari Islands, north-westof Misol (pl. 15, p. 172). Thereis a model of a largeplank boat fromGowa, South Celebes,in the Leiden Museum,with a double outrigger,five booms witha directattachment to each float; of these the foreand aft booms are insertedinto the float,while the threemiddle ones are lashed to the undersideof the float. It wouldnot be surprisingif this were a model of a craftcoming from the Sulu area (p. 114). There are usuallythree booms at Buton,an islandoff the south-eastpoint of Celebes (Friederici, II, p. 235). Professor S. J. Hicksonhas presentedto the CambridgeMuseum several small modelsof cere- monialsakit canoes fromNanusa Island, north-eastof Celebes; theyhave eithertwo or three booms which are pegged' on to the floats,or the outriggermay be absent (1893, Fig., p. 290). This island is just on the above-mentionedborder line. In spite of these exceptionsthe canoes of Celebes appear predominantlyto have but two booms. Morethan two boomsusually occur in the SangirIslands (Miller,p. 244). In the Sulu Islands small canoes may have but two booms (Guillemard,1889, p. 206; G., Nos. 180, 217), but usually thereare three (Fig. 7) (Wilkes,V, p. 333; Buxbidge,p. 225; G., No. 4) or four (Guillemard,1889, p. 192; G., Nos. 115, 142,

FIG. 7.-CANOE WITH DOUBLE OUTRIGGER, THREE BOOMS, AND DIRECT LASHED ATTACHMENT, SULU (Photo G., No. 4).

I do not regard the pegs as being significant; they are probably merelya labour-saving device.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 81

180, 192, 211; model of a " piratical prahu," dapang, Edinburgh Museum). In these canoes the foreand aft booms are always straight,whereas the centralbooms are downwardlycurved at their ends (Fig. 7). Pritchett(p. 183) illustrates a " Pirate craftoff north point of Borneo" witha double outriggerconsisting of three downwardlycurved booms (Fig. 8); it was probablya Sulu vessel.

FIG. 8.-PIRATE CRAFT OFF NORTH POINT OF BORNEO (AFTER PRITCHETT, p. 183).

In the Philippinesthree booms may occur at Manila and on Lake Lanao in Mindanao (Vojnich,pp. 378, 383). In numerouspictures of the East Indian seas givenby old Dutch voyagersthere are depicted large sailing war-vesselswith massive outriggerspossessing several (usually three) booms,on or underwhich are fasteneda variable numberof longi- tudinal planks (i.e., parallel to the hull) which are used as seats by the paddlers. I shall referlater to these craftand to the ancientIndo-Javanese vessels which had at least three booms to their outriggers. Weule figures(P1. 112, Fig. 7) a double outrigger" Boot von den Molukken,"but gives no furtherparticulars; it has three booms and what may be a Moluccanattachment, a board is lashed to the underside of the booms midwaybetween the hull and each float,on whichthree men sit and paddle. ThroughoutPolynesia, with but fewexceptions, the canoes have two outrigger- booms. Two straightbooms are found in the Pelew, Marianne,and typicallyin the CarolineGroups; characteristicof the MarshallGroup are two straightbooms, the end of each of whichis supportedby a straightstick which is insertedinto the floatand apparentlyalso into the boom, on each side of the booms are threecurved VOL. L.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 82 A. C. HADDON.-The, Outriggersof Indonesian Canoes.

booms,the ends of whichare lashed to the float(Model, Salem Museum; Alexander, pl. 36, pp. 805, 806), a variant fromNonuti, GilbertIslands, is shown in Fig. 29 see also p. 124. Equally characteristicis the presenceof three(or four)out-rigger- booms in Melanesia,but in New Guinea there is considerablediversity. In all of theseareas there is a singleoutrigger, with the exceptionsnoted on pp. 79, 122; in those cases wherethe outriggeris double thereare but two booms. The onlyinstances known to me in Indonesiaof but one outrigger-boomare those associated with a single outriggerin a toy boat fromBorneo (Fig. 32), and in the case recordedby Hornellfrom North Java (1919, Fig. 1), and the modelfrom Manila (p. 114, Figs. 30, 31) in whichthe outriggeris double. ProfessorJ. StanleyGardiner has givento the CambridgeMuseum a toy canoe, abbuodi,from the Maldive Islands. It has a singleoutrigger and only one boom,the end of whichis morticedinto the float. He informsme that boys frequentlyuse canoes withone, two, or threebooms. The men's canoes have no outrigger,but the old people say that outriggercanoes were formerlyused extensively; one man called them " rafts." There is a model in the Peabody Museum,Salem, Mass., of a canoe fromRuk, CarolineGroup, with a long singleboom on one side only whichis strengthenedby two horizontalconvergent sticks from the foreand aft quartersof the bull; thereare two centralparallel short ridges on the canoe-shapedfloat, the boom restsin a notch in the innerone and is insertedinto a hole in the outerone. Muller (1917, I, Fig. 295, p. 199) gives an illustrationof a toy sailingcanoe, tatareg,from Yap in the same group; it is somethingsimilar to the precedingone, but thereis an obscureindirect attachment. The play-boat figuredby Kriimer (1906, p. 291) fromthe Gilbert Islands appears to be very similar.

THE ATTACHMENTOF THE BOOMS TO THE HULL. In the majorityof cases the booms rest on the gunwales,especially when the hull consistssolely of a dug-out. In the Andamans,however, the boomspass through holes in both sides of the dug-out(Fig. 11, A, D). It not infrequentlyhappens that a length of sapling or bamboo is lasbed to the upperborder of the dug-outand the boomsrest on thisgunwale spar, as it may be termed. The pbject of this spar is obviouslyto protectthe edges of the hull; it is widelydistributed in Oceania. In the Nicobars,according to A. R. BrownMS., the boomsrest on the gunwales and a gunwalespar is fastenedover themalong the lengthof the canoe (Fig. 10, B), but Svoboda (VI, P1. 1) figuresthe ordinaryarrangement. Sometimesthe booms lodge in notchesin the gunwale,or the gunwalemay be locally raised at the spots which support the booms; the upper edges of these protuberancesmay be notched.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 83

THE FLOAT. The floatusually consistsof a singlelog of wood or piece of bamboo. When not otherwisestated it may be assumedthat the floatis single,but occasionallytwo or morebamboos (or pieces of wood) are employed,in whichcase it may be termeda double,treble, or multiplefloat. The float,when made of bamboo, has its ends cut square, but when made of wood it is usuallytrimmed; the foreend only or both ends may be pointed,or the foreend and occasionallyboth ends may have an upwardcurve.

THE ATTACHMENTS BETWEEN THE BOOMS AND THE FLOAT AND THEIR DISTRIBUTION.' Thereare variousmethods in the attachment betweenthe booms and the float. The main groups are: A.-Direct. 1. Inserted.-The ends of both booms are insertedinto the float. This appears to be very rare, if it actually does occur in practice,in Indonesia. Accordingto Folkard's drawing (p. 480) the Sumatranjellore and ballellanghave booms with downwardlycurved ends whichappear to be insertedinto the floats; unfortunately the drawingsin his book are on too small a scale forthe details of attachmentto be reliable. The toy boats of Sarawak (Fig. 32) have this method. The Balinese attachment(p. 88) reallybelongs here. 2. Lashed.-The ends of all the booms are lashed to the float(Figs. 1, 2, 5, 7). This type is widelyspread throughoutIndonesia. It has been noted at Mentawei (Rosenberg,P1. XVIII, Fig. 9); Engano (model,Leiden); Palembangin Sumatra model, Amsterdam); Madura (model, Amsterdam; model, Leiden, in this case each floatconsists of two bamboos betweenwhich is a small peg whichprojects on the under and upper surfacesof the boom, evidentlyto renderthe lashing more secure); Java (L., I, p. 35b); Bali (Fr., II, p. 235, Fig. 1, on a dyukuncanoe); Sumba or Sandalwood Island (model,Amsterdam); Lomblen,between Flores and Timor (model,Leiden); Baba, west of Tenimber(Pfluiger, p. 145, no description); freshwatercreek of Totoat, Kei Islands (Langen,p. 52, poor figure,no description); Lintschotusillustrates a sailing vessel apparentlyof this type with the inscription " Navium quibus Bantani utunter" (Tertia pars, DCI, XXIIXh), but the booms may have been insertedinto the float; Lake Wakollo or Wakoholo,Buru (Forbes, P1. p. 405, no description; K. Martin,1894, p. 329, P1. XLV, with three slightly curved booms); Batjan (G., No. 336); Ternate (L., II, Pls. 14, 15; Kiikenthal, P1. 7, Fig. 13, obscure,no description);Lake Galela in NorthHalmahera (Kiikenthal, p. 172); Makassar (Valentijn,No. 23, p. 136); Gowa in South Celebes (model, Leiden); Paloppo and Libukang in the Gulf of Boni, Celebes (P. and F. Sarasin, 1 Map, p. 71. The distributionoutside of Indonesiais givenlater, p. 124. G 2

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 84 A. C. HADDON.-The Outriggersof Indonesian Canoes.

1905,I, Fig. 53; II, Figs. 62, 63); Ussu on the MaliliRiver and Lake Matanna,at the north-eastend of the Gulfof Boni and Paloppo (Grubauer,Figs. 15, 38, 39, 108); Kema, North Celebes (G., No. 62); Sangir (model, Amsterdam,single outrigger withtwo booms and an outlayer,Fig. 5); Talaut Islands (" Geisterkahn" or sakit canoe, Dresden,Meyer and Richter,P1. I, Fig. 10); Sulu Islands (Wilkes,Vol. V., p. 333; Burbidge,p. 225; Savage Landor, II, pp. 2, 12; Muller,p. 244, Fig. 55; Guillemard,1889, pp. 192, 206 ; G., Nos. 4, 142, 180, 211, 217) ; Cagayanes Group (Savage Landor, I, p. 228); Manila and districtof Lake Lanao, Mindanao (Vojnich, pp. 378, 383); Pasig River,Manila (p. 114, Fig. 30); Zamboanga and Zebu (Muller, p. 244, Fig. 57); the " Pirate craftoff north point of Borneo" (Fig. 8) (Pritchett, p. 183) appears to belonghere. In a largecanoe at Amboinafigured by Valentijn(No. XXX, p. 124) the booms seem to be lashed directlyto the float,they have a zigzag appearance and support planksfor paddlers, but the illustrationis not convincing. 3. MixedDirect Attachment.-A model of a fishingboat, sampan,from Panarukan, Madura Strait,in the RotterdamMuseum, has two booms,one of whichis straight and tied to the float,while the otheris downwardlycurved and insertedinto the float. Practicallythe same arrangementis shown in a model from" Java "' in the Am- ,sterdamMuseum (Fig. 9). In the Leiden Museumare two modelsfrom Madura, in whichthe foreboom is straightwith its endslashed to the floats,while the aft boom is shortand straightand has insertedinto each end a straightspar witha downwardly

FIG. 9.-MODEL WITH DOUBLE OUTRIGGER AND MIXED DIRECT ATTACHMENT, JAVA (AMSTERDAM).

curved end which is inserted into the float; these booms are probably intended to represent a sedek(p. 89), and if so this is an example of a mixeddirect and Balinese attachment.I think in all these cases it is the fore boom which is lashed to and the aft boom which is inserted into the float. A model of a large plank boat in the Leiden Museum from Gowa, South Celebes, has five booms of which the fore and aft booms are inserted into the float, while the three middle ones are lashed to the underside of the float (pp. 80, 106, 113, 114).

B.-Indirect. (a) Attachmentinserted into theFloat. Stick Attachment.-Attachmentswhich consist of one, two, or more sticks,one end of each stick is lashed to the boom while the otheiris insertedinto the float.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 85

T-hesticks may be vertical,in whichcase theremay be only a singlestick, or there may be two or moresticks, either on one or on both sides of the boom; the sticks may be irregularlyoblique, in oblique parallel pairs or a pair of oblique sticksmay convergeover the boom; a pair of sticks may cross under the boom, undercrossed, in whichcase the boom typicallyrests on the crossing,or the crossingmay take place overthe boom,overcrossed. In the outriggercanoe, due, of the Nicobars the attachmentconsists typically of a double set of threesticks, heneme, which are insertedinto thefloat, hentaha, and lashed to the boom, deja due, in such a way that two sticks generallycross each otherbelow the boom, whilethe thirdmay be verticalor oblique, and may be fore or aft of the boom (Fig. 10, C, D); occasionallya pair of sticksconverge over the boom,the thirdbeing more oblique (Fig. 10, F). The two sets ofthree sticks diverge fromthe median line of the float (Fig. 10, A). Sometimesthere is also a' central pair of undercrossedsticks, which is insertedvertically between these two, as in

C~~~~

- ~F FIG. 10.-STICK ATTACHMENTS, NICOBARS: A-C (A. R. BROWN MS.) C, REPRESENTS THE ARRANGE- MENT OF THE STICKS; IN A, ONLY TWO OF THE SIX STICKS ARE SHOWN; D, E, MODELS (OXFORD); F, MODEL (EDINBURGH). modelsin the OxfordMuseum (Fig. 10, E). In a model in the EdinburghMuseuma one set consistsof two sticks whichare almost parallel and convergeslightly over the boom and a thirdoblique stick (Fig. 10, F). The native names are taken fromSvoboda, who does not describethe arrange- mentof the sticks,nor are his figuresat all clear (VI, P1. I, Figs. 11, 17); he also gives an illustration(V, p. 193) of a toy canoe in whichboth booms pass throughthe hull and rest on the float and are tied to an adjacent shortpeg whichis inserted into the float. A good illustrationof a model of a sailingcanoe, whichshows the details of the sticks,is given by Man (XI., P1. XXIV), later (XV, pp. 436, 449) he alludes to Nicobaresecanoes withoutsaying anything about outriggers. Kloss (pp. 53, 79) givesshort accounts of the canoes,ap, and two excellentplates (pp. 80, 154);

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 86 A. C. HADDON.-The Outriggersof Indonesian Canoes.

the latter shows the attachmentsvery well, as does the plate, p. 345, Journ. Anthrop.Inst., VI, 1877,cf. p. 209. The Andamaneseattachment consists most frequentlyof one set of two under- crossed sticks and one vertical stick, which may be on either side of the boom (Fig. 11, B). Thisarrangement occurs in one attachmentin a modelin the Cambridge Museum,while in the othertwo attachmentstwo stickscross over the boom and a thirdvertical stick is present(Fig. 11, D). In a model in the HornimanMuseum the threebooms each have an attachmentof onlytwo stickswhich converge over the boom. Modelsin the BritishMuseum, Oxford, and EdinburghMuseums, have three booms with a singlepair of oblique stickswhich cross underthe boom (Fig. 11, E).

A C

RIG. 11.-STICK ATTACHMENTS, ANDAMANS: A, B, GREAT ANDAMAN, C, LITTLE ANDAMAN (A. R. BROWN MS.); D, MODEL (CAMBRIDGE); E, MODEL (BRITISH MUSEUM).

In the Little Andaman(A. R. BrownMS.) thereare threevertical sticks, two on one side of the boom and one on the other. There is, however,some variationin the arrangementin the various models,even in the attachmentsof the several booms in the same model; forexample, in the Halifax Museum,a model of a large canoe, roko,has threebooms-the attachmentof two is typical,while in the thirdthe boom passes betweena pair of crossedsticks and the verticalstick, but this is probably due to carelessworkmanship. There is frequentlyin addition a couple of lashings (Fig. 11, A, D, E) or a singleone, of ratan fromthe boom to the floatto renderthe attachmentmore secure. A good photographof an outriggercanoe, chd-rigma,and of the largerkind without outriggers, gi lyanga,is givenby Man (XII, p. 116,P1. VII).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 87

With these exceptions,a stick attachmentof this kind has not been recorded forIndonesia, though, as we shall see (p. 125), it is very commonin Oceania. Rod Attachment.-Thisattachment, which consists of a moreor less verticalrod or stickinserted into the floatat its lowerend, has not hithertobeen recordedfrom Indonesia as an attachmentfor all of the booms of a canoe. That it may have occurredthere is possible,as it is foundin some mixedtypes (pp. 92, 96). To avoid confusionwith the above-mentionedstick attachmentthis may convenientlybe termeda " rod attachment." SpikeAttachment (pp. 79, 127). Y-shapedStick Attachment (p. 127). Board Attachment.-Hornellhas recentlydescribed a form of attachmentto wbich the term " board " may be applied (1919, No. 55). It consistsof a short, broad and relativelythin piece of wood; its lower end is insertedinto the float whilethe boom passes throughits upperend; in these two respectsit resemblesthe morestick-like attachment of the East Africancanoes (pp. 79, 128; iladdon, 1918, No. 29). This attachmenthas been recordedonly fromthe northcoast of Middle Java, whereit is associated witha singleoutrigger and a singleboom. It seemsto be related,however, to an attachmentfigured by Nicolas (1601, II, Pls. 14, 15, 17) on trading-and war-vesselsat Ternate (p. 110) and by De Bry (1601, V, PIs. XII, XVI) on similarcraft from the Moluccas. At that time the larger boats of the Moluccas had a double outriggerof three stout straightbooms, each of whichwas supportedin a deep notchin a shortboard which was apparentlyinserted into the float(Fig. 12 B). I proposeto referto thistype as a Y-boardattachment. Valentijn

- :

t =~~~~~~~~~~0~

FIG. 12. CORACORA A," THE DOUBLE OUTRIGGER HAS A FLOAT CONSISTING OF TWO PLANKS; B, "THE KARKOLLA OF THE KING OF TERNATE," WITH A Y-BOARD ATTACHMENT; ABOVE

THIS IS A MADURA WAR SHIP (DE BRY; AND NICOLAS, 1601, pl. 9, p. llb.)

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 88 A. C. HADDON.-The Outriggersof Indonesian Canoes.

(p. 363) showsthese boards as also lashed to the floatson Moluccanvessels. A thick longitudinalspar or board ran across the ends of the booms, presumablyto keep themin place, this was frequentlyutilized as a seat for paddlers. In one coracora there was apparently a straight Halmaheran attachment near each end of the longitudinalspar. Generallyone or two planks are fastenedtransversely to the boomsupon whichmen sit to paddle. In an illustrationentitled " De Cora-cora van Titaway" (Fig. 13) there are six outriggerbooms which are supportedby as

D., CORA -CORA, VAN TIAWAY

FIo. 13 DE CORA-CORA VAN TITAWAY " (VALENTIJN, No. XLII, p. 184). many short,thick, squared bars of wood whichappear to be insertedinto the float, a plank on whicheight men sit to paddle replacesthe longitudinalspar, there are four othersimilar planks on the lowermostof whichare also paddlers. BalineseAttachmnt. We may adopt Hornell'sterm for the type of attachment whichconsists of a straightor slightlycurved spar one end of whichis spliced,pegged and lashed to each end of both of the shortstraight booms, the otherbeing inserted into the floatand reinforcedby lashing(1919, No. 55). This type was firstdescribed and figuredby Friederici(II, p. 235, Fig. 2) (Fig. 14 A), who gives the Bali names for

A

FIG. 14.-BALINESE ATTACHMENT: A, BALI (FLIEDERICI, , fig. 2); B, MODEL FROM BANJUWANG!, E. JAVA (EDINBURGH).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndowesianCanoes. 89

boom,brayunan, " intermediatepiece," sedek,and float,kater. Hornellgives a good photograph(1919, P1. G., A) of a Lombokcanoe, and figuresdetails of the sedekfrom Boleleng,Bali. Thereis a model of a sailingboat, dukong,from Banjtiwangi, Java, on the Bali Strait,in the EdinburghMuseum with this attachment(Fig. 14 B). Hornellsuggests that this is a device forextending the boom downwardsso as to save the necessityfor searching for a naturally-bentpiece of wood withwhich to constructan outrigger-boomof the requiredform. As I have stated (p. 73), the sedekmay be regardedas one formof the " boom-prolongation."If the sedekwere a formof Halmaheranattachment one would expectit to be onlylashed to the float, whereasit is insertedinto it, as it appropriatelywould be if it werederived froma directinserted attachment; the upper end is virtuallyan integralpart of the boom and not a sparlashed on to it. As the sedekis an added piece,this form of attachment is strictlyspeaking " indirect,"but if it be regardedmerely as a prolongationof the boom it shouldbe describedas a " directinserted attachment."

(b) Attachmenttied to theFloat. Moluccan.-Tbe typicalMoluccan attachment is formedby a U-shaped piece of ratan,the horns of whichare lashed to one side of the freeend of each ofthe booms, and the base is lashed to the float(Figs. 3, 15).

A B C

FIG. 15.-MOLUCCAN ATTACHMENTS: A, BATJAN; B, AMBOINA; C, BANDA (A, C, FROM PHOTOGRAPHS, G., 334, 161; B, AFTER FRIEDERICI, ii, fig. 7a).

It has been noted from models at Wetta (Riedel, Pls. XLI, Fig. 12; XLIII, Fig. 8, Baessler,p. 78, P1. VII, Fig. 4); Batu mera in the south-eastislands of the Kei Group (model,Amsterdam); Buru (Riedel, frommodel, P1. I; Fr., "here and there," III, p. 161); Amboina (K. Martin,1894, p. 232; Fr., II, Fig. 7a, p. 237; model,Leiden, on aft side of all the booms); Uliassers (Fr., III, p. 161); Ceram (Fr., III, p. 161, " with the exceptionof a few places on the northcoast"; South Ceram,K. Martin,1894, p. 232, Pls. XII, XXVIII, Fig. 15); Ombi,north of Ceram (Fr., II, p. 239); Batjan (Fr., II, p. 239, " predominates"; G., Nos. 328, 331, 334, 336); Ternate(Kiikenthal, P1. 5,Fig. 8; Fr.,II, Fig. 11, p. 240," greatlypredominates"; G., No. 327); northernHalmahera (" occasionally" at Ake-Selaka, Tobelo and Galela, Fr., II, pp. 240, 242, Figs. 11,27a; III, p. 161); Buton (Fr., II, Fig. 3, usually withthree booms and threebamboos to formthe float,p. 235). Othervarieties of this type are the 0-shaped attachmentwhich predominates at Amboina (Fr., II, p. 237; Pflilger,p. 131; models,Leiden), and the 6-shaped attachmentwhich i's the commonform at Banda (Fr., II, p. 237; G., No. 161).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 90 A. C. HADDON.-The Outriggersof Indonesian Canoes.

The Moluccan attachmentsof boats between Selang and Batjan figuredby Forrest(pl. 4, p. 82, pl. 5, p. 86) are not veryclear; the formeris a " MoluccaCoro- coro," and has threebooms, a tripodmast and a lyretanjong sail. Halmaheran.-This attachmentconsists of a variously shaped spar which is lashed above to a boom and below to the float(Figs. 4, 6, 16, 17). In orderto preventconfusion with the " stick" attachment,which is inserted into the float,I proposeto referto this elementas a "spar," even whenit is a simple rod or stick. This is the " oblique " or " elbow-stanchiontype " of Hornell. The term " Halmaheira-Verbindung,"like that of "Molukken-Verbindung,"was introducedby Friederici(II, p. 239). The simplestcondition consists of a straightspar whichmay be vertical,as in a Batjan canoe (Fig. 16 F; G., 336) and in a sailingvessel, bero, at the TenimberIslands (Riedel, P1. XXVII, Fig. 9, I assume that the spar is tied to and not insertedinto

A

(C) ~

FIG. 16.-HALMAHERAN ATTACHMENTS: A, BUTON; B, C, ARKE SELAKA, AND D, WEDA BAY, HALMAHERA (FRIEDERICI II, pp. 235 f.,figs. 4, 22,17, 27); E, MODEL FROM TALAUT (AMSTER- DAM); F, BATCHAN (PHOTO G., 336).

the floatin this case); or oblique, as in a canoe on the Bay of Bara, northcoast of Buru (K. Martin,1903, P1. XIV) and Banda (G., 346). The attachmentof the central boomsof the Sangir canoe (Fig. 2, D) may be a variant. The sparmay be forkedand practicallyhorizontal (model from Tenimber, Leiden), p. 76 and Fig. 4; usually it is more or less vertical,Batjan (G., Nos. 328, 331), Misol (G., No. 267) or oblique,North Ceram (K. Martin,1894, P1. XXVIII, Fig. 16).

FIG. 17.-CANOE WITH DOUBLE OUTRIGGER, HALMAHERAN ATTACHMENT, AND AN INNER AND AN OUTER LONGITUDINAL SPAR. BATJAN (FROM PHOTO G., 331).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 91

Mostfrequently the spars are angled,elbowed, or bentin variousways (Fig. 16). If the main stem of a forkedspar werecut offimmediately above the forkan angled spar would result. The stem may be straightand the upper part bowed (Fig. 6); the spar may have a slightsigmoid flexure, which may be so slightas to be almost straight. Occasionallythe spar is sharplybent, with the ends pointingdownwards, as in a Buton canoe (Fr., II, p. 235, Fig. 4), or withthe ends pointingupwards as in a model fromTalaut in the AmsterdamMuseum (Figs. 16 A and E). These variations,so faras my data go, do not appear to be significantas regards formor distribution,since nea'rlyevery varietyoccurs on the coasts of or on the islands immediatelyadjacent to Celebes, and elsewhereseveral varietiesare found pnthe same spot; sometimestwo varietiesmay be foundon a singlecanoe. It occurs at Lombok (" common,"Hornell, p. 99); Timor (photo.); ? Baba (Pfluiger,p. 147); Tenimber(Riedel, model,P1. XXVII, Fig. 9; model, Leiden); Dobbo, Aru Islands (O. Warburgphoto. in Krieger,P1. 29); Buru (K. Martin,1894, P1. XLVI, 1903,P1. XIV; Hornell,p. 99); Banda and Ambon(" occasionally,".Fr., II, p. 239); Ceram,north coast (K. Martin,1894, p. 232, P1. XXVIII, Fig. 16; Fr., II, p. 239; Hornell,p. 99); Misol (G., No. 267); Ombi (Obi) (Fr., II, p. 239, " commontype"; Hornell,p. 99); Batjan (" a few," Fr., p. 239; G., Nos. 328, 331, 336); Tidor (K. Martin,1894, p. 233); Ternate(model, Leiden); Halmahera Weda Bay (Fr., II, p. 242, Fig. 27), Ake Selaka (Fr., pp. 240, 243, Figs. 17, 22, 23), Patani and. Buli (two booms, oftenwith a centralfalse-boom, Fr., p. 242), Tobelo and Galela (" greatlypredominates," Fr., pp. 240, 242); Xulla, or Sula (K. Martin, 1894, p. 233; Fr., III, p. 161; Hornell,p. 99); Buton (Fr., pp. 235, 239, Fig. 4); Celebes: Makassar (R. Martin,'1894, p. 233); Gowa (model, Leiden); Konaweha River, South-east Celebes (Sarasin, I, p. 376, photo, but no description); Malili River,at the north-eastcorner of the head ofthe Gulfof Boni (Grubauer,Figs. 8, 10); Gulfof Gorontalo or Tomini: Pogoyama(G., No. 242), Todjo, Gulfof Tomini (Adriani and Kruyt, No. 9, Chap. 5); Minahassa (Dumont d'Urville L'Astrolabe, Atlas, Pls. 192, 204), Kema (G., No. 232), Limbe Island (Guillemard,P1. p. 325); Menado (Hornell,P1. G, B, and Fig. 3; models,Leiden); Tontoli (Toli-Toli), North-west Celebes (Pfluiger,p. 85). Althoughthey are nominallybeyond the scope of this paper, the outrigger canoes of North-westNew Guinea must be alluded to as they differfrom those of otherparts of New Guinea and undoubtedlyare of directIndonesian origin, and cdn be perfectlymatched by a type fromWeda Bay in Halmahera (Fig. 6). So far as I can gather,but one type of outriggerextends from Skroe to Waigiu and possibly to Manukwari(Dorei) at the westerlyopening of GeelvinkBay, whereand further east it is replacedby otherattachments. At Skroe (a port foundedby the Dutch in 1899 on the northshore of Telok Kampauer, i.e., on the south side of Onin Peninsula),judging from Pfluiger's small and indistinctphotograph (p. 171), the double outriggerhas fourbooms, the two fore

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 92 A. C. HADDON.-The Outriggersof Indonesian Canoes. and the two aft of whichare nearertogether than are the two centralones; there is a Halmaheranattachment of the Waigiu type. Apparently,according to Pfluiger's photographs(pp. 174, 175) a similartype occursat Sekar (Segaar) on the southshore of Telok Berow (Berou) or McClureInlet, i.e., on the northside of Onin Peninsula; Kreigersays they have veryhigh boats with outriggers(p. 385). Dr. Guillemard's photographs(G., Nos. 305, 427, and 1889, p. 373) of canoes in ChabrolBay, Waigiu (Waigeu6),shows the same type of canoe as that recordedby Friederici(II, p. 248, Figs. 29, 30; Hornell,p. 99) at Saonek, an islet near the south coast of Waigiu and amongthe Sorong(Soron) people on the island of Dom, who removedthere in 1865 fromSorong, a villagenear Cape Spencer,or Kaap Noi, on the mainlandof New Guinea (Fr., II, p. 248, Fig. 28; Hornell,p. 99). The dug-outor plank-builtcanoes have double outriggers. In the interiorof the hull are transversestruts and lateral longitudinalspars as at Manukwariand Japen (Jobi)in GeelvinkBay to the east and at Halmaherato the west. The sides of the largercanoes are heightenedby super- imposedwash-strakes, and not by gabbagabbaas in furthereast. The fourbooms of the outriggerstretch across the wash-strakes,to whichthey are lashed by means of ledges(" Leisten" or patnati),but in small canoes they rest directlyon the edges of the dug-out. The two floatsmay be shorterthan the hall, or, as in Sorong,like sledge-runners,and, as in Tahiti, run far forward,the last attachmentspar being near its aft end. There is a Halmaheranattachment of a long fairlystraight spar with a bentend, sometimes it is an elbowedspar; an innerlongitudinal spar passes over the fourbooms and underneaththe angle of the spars. Friedericifigures an attachmentat Saonek with an additional outer longitudinalspar, the floatin this instanceis composedof two bamboos instead of the usual single one. The larger crafthave a platformwith side rails and an atap roof. On the booms,on both sides of the canoe, thereare usually forkedsupports for gear, whichhave a crescenticor otherform; on one sidelies theunshipped triangle-mast and on the otherthe rolled-up sail (Fr., II, p. 248).

C.-Mixed Attachments. In the foregoingaccounts the attachmentsare similaron all the booms of a canoe. I now proceedto give examplesof mixedattachments. The mixeddirect attachment and mixeddirect and Balineseattachment have already been described(p. 84). Mixed Directand Rod Attachment.-Pricbett(p. 175) gives a drawing(Fig. 18) of a sailingcanoe, sukung,from Probolingo, Madura Strait,in whichthe foreboom appearsto be a boom-prolongationwhich is lashed to the float,or it may be that the boom is in one piece, but its diametermarkedly varies in parts. The same applies to the aftboom exceptthat the boom-prolongationis thickerand is upwardlycurved with a swollenend. this is connectedwith the floatby means of a rod whichmay be lashedto the boom or may pass throughit; the lowerend appears to be inserted

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 93 into the float,but the drawingis not decisive as regardsthese two points. In referringto the "outriggersupports "lhe says, "The one forwardbeing low downand

FIG. 18.-FISHING BOAT, sukunzg,WITH MIXED DIRECT AND ROD ATTACHMENT, PROBOLINGO, MADURA STRAIT (PRITCHETT, p. 175). thataft curving up pronouncedly,to allow the washto pass underfreely when the vesselis at herhigh speed " (p. 174). A model(Fig. 19) of a sekongin the Rotterdan-i Museum from Pasuruan, in Madura Strait,seems to clearup thepoints that are doubtfulin Pritchett'sdrawing. Both ofthe short booms have a boom-prolongation,thefore one is lashedto thefloat, the

FIG. 19. MODEL OF A sekong WITH MIXED DIRECT ANiD ROD ATTACHMENT, PASURAN, E. JAVA (ROTTERDAM). freeend of the aft one is expaindedand decoratedwith fret carving, a rod passes throughthis boom-prolongation and is insertedinto the float, a lashingalso connects theboom and the float at thisspot.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 94 A. C. HADDON.-The Outriggersof Indonesian Canoes.

In a sketchby Muller(1912, p. 244, Fig. 22) of a canoe fromMadura (Fig. 20), the foreboom is a doublybent yoke-shapedbar, the ends of whichare lashed to the floats; the aft boom, or boom-prolongation,is stronglycurved upwardly,being almost U-shaped, each divergentlimb being connectedby means of a vertical T-

+ 0

FIG. 20.-SKETCH OF A MIXED DIRECT AND ROD ATTACHMENT, MADURA (FROM MULLER, 1912, fig. 22).

shaped rod withthe float; the transverseupper end of the rod is presumablylashed to the boom and probablyits lower end is insertedinto the float. Thus in all the main points the outriggeragrees with that of the Probolingoboat. The same authorgives a sketch(Fig. 21) of a canoe fromthe Bawean Islands; the foreboom is straight,how it is attached to the floatis not evident,but it is a direct attachment. The author states that at the stern the bamboo floats are suspendedwithout spars (Auslegerstange)from an elastic bent rod whichis concave above (p. 244). If, as his sketchindicates, this rod is all in one piece,it mustbe made of bent ratan, as it is inconceivablethat the wholeapparatus could be made out of

FIG. 21.-SKETCH OF A MIXED DIRECT AND ROD ATTACHMENT, BAWEAN ISLANDS (FROM MULLER, 1912, fig. 23).

a single piece of wood. In either case the formrequires explanation; we may thereforesuppose that it consistsof the characteristicallyupwardly curved boom or boom-prolongationto each end of whicha verticalrod is fastened,the lowerend of whichis apparentlyinserted into the boom; if this be so, it falls in with the previousexamples. Muller states that he made these sketcheswhilst on board a steamer. Mixed Direct and Halmaheran Attachment.-Inthe region embracingNorth Celebes, Banka, the Talaut (includingSangir) and Tulur Groups canoes have two outriggersand manyof them have but two booms.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. 0. HADDON.-The Outriggersof IndonestanCanoes. 95

Friedericisays: " In Minahassa there are outriggerboats whose foreboom is curved and-exactly as in Bali, Bugi, Makassar, at the Mariannes,in Tahiti and at the Marquesas-is fastened directlyto the float,while the aft straightboom is connectedby means of an S-shaped spar with the float,sema. Accordingto the descriptionof Graafland(Minahassa, II, pp. 404, 405) this mightbe consideredas a Moluccanattachment, but it is really,as Dumontd'Urville's drawings (Atlas, P1. 234, Figs. 1, 2) prove, the Halmaheran attachment" (III, p. 161). Hicksongives a figureof a model of a sailingdug-out, londi, from Talisse in the Banka Strait(Fig. 1, p. 22). The foreboom is stronglycurved and its endsare lashed to the floatswhich are generallymade of two or threepieces of thickbamboo firmly lashedtogether. The aftboom is straightand its ends are connectedwith the floats

FIG. 22.-SAILING CANOE, londi,WITH MIXED DIRECT AND HALMAHERAN ATTACHMENT TALISSE ISLAND, N.E. CELEBES (HICKSON). by means of an S-shaped Halmaheranattachment (Fig. 22). In some dug-outsat Kema, slightlyto the south of Banka Strait,the fore-boomis downwardlycurved witha directlashed attachment,and the shorterstraight aft boom is attachedto the floatby a bent Halmaheranspar (G., No. 67). A modelcanoe from the Talaut Islands in theAmsterdam Museum has two straight booms,of whichthe fore one has a directtied attachmentand the aft one has a bent or bowed spar attachment,but in this case the freeends of the spar are lashed to the boom and the bend to the float,it thus bears some resemblanceto the U - Moluccanattachment (Fig. 16 E). It may be a spar ofthis kindto whichGraafland refers. In Sangir,according to Muller(1912, f.n., p. 244), the two outer (foreand aft) booms of the double outriggerand their boom-sparsbend downwardsand are

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 96 A. C. HADDON.-The Outriggersof Indonesian Canoes.

connecteddirectly with the floats,whereas the centralbooms have the structureshown in Fig. 2, D, and have a l-shaped attachmentspar, the lowerend of whichis lashed to the double floats. Thereis a veryremarkable model in the OxfordMuseum, which was collectedby Capt. J. P. Maclear of the " Challenger." The dug-outhas notchedends, the upper part beingprolonged into a shortupwardly slanting beak, the lowerpart of the bow is producedinto a long spur whichrises in a gentlecurve. There are two floatson each side,each of whichhas theordinary pair of booms; the two forebooms lie very close together,as do the two aft booms. The fore booms have a direct lashed attachment. The outerfloat is attachedto its aft boom witha bowed Halmaheran attachment,while the inneris attachedto its aft boom by a '-o-shapedHalmaheran attachment. The forkson the booms have threebranches, one long and low down (like the browtine of a deer's antler),the othertwo shortand terminal. The fore forksare lashed to the stouterand hinderof the two forebooms, i.e., the oneattached to the innerfloat; a spar rests on the lowestbranch of the forkand is lashed in its middleto the socketof the mast. The innerboom is securedby two lashingswithin the hull of the canoe to a cross-bar,the ends of whichare kept in place by passing belowprojections left in the innersides of the hull. The mast is steppedin a socket consistingof a section of bamboo, the upper end of wiich is lashed to the hinder foreboom, while the lowerend is steadiedby a cross-barpassing through it, the ends of which abut against the inner sides of the hull. I do not know of any other exampleof two floatson each side each withits own attachments,the double,treble or multiplefloats previously noted (p. 83, figs.1, 2 D, 15 B), are treatedas ifthey were simplefloats. Therecan be no doubt that the specimencame fromthe NorthCelebes area,possibly from the Talaut Group,or possiblyfrom the Nanusa Islands as the ends of the canoe resemblethose of the sakitcanoes describedby Hickson(cf. pp. 77, 113). Mixed Rod and HalmaheranAttachment.-A photograpph (G., No. 232) at Ken-a, Minalhassa,North-east Celebes, shows a canoe with attachmentswhich are distinct fromany other knownto me. At both ends of the canoe there is a framework consistingof a shortstraight boom, on whichis an equallyshort upwardly curved spar, or boom-spar,these are braced by two verticalsticks. At what is presumablythe foreend of the canoe thereis a verticalrod whichis fastenedto the ends of its boom

FIG. 23.-MIXED ROD AND HALMAHERAN ATTACHMENT KEMA, N.E. CELEBES (PHOTO G., 232).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-T he Outriggersof Indonesian Canoes. 97 and curved boom-sparand appears to be insertedinto the float. At the otherend the rod is replacedby a bent spar, whichappears to be a true Halmaheranattach- ment,as it is lashed to the float. The firstof these attachments somewhat resembles the aft attachmentof the Bawean canoe (Fig. 21). This Kema type can easily be resolvedinto the Sangirtype (Fig. 2, D) ifthe boomsand upwardlycurved boom-spars of the latter were greatlyshortened and the oblique sticks or stays were placed vertically. The straightrod also correspondsfairly closely with the l-shaped spar of the Sangirtype, except perhapsfor its attachmentto the float.

NOTES ON THE CHARACTERISTIC OUTRIGGER CANOES OF THE MAIN DISTRICTS OF INDONESIA.

The outriggercanoes of the Andamansand Nicobarshave been sufficientlydealt with on pp. 79, 80, 85, 86. The main point to rememberis that in both groups there is an insertedstick attachment,but in the Andamans the single outrigger has severalbooms, while in the Nicobarsit has but two. There is a markedabsence of outriggersfrom the greaterpart of Sumatra and Java and fromthe whole of Borneo,which is evidentlydue to a knowledgeof the art ofbuilding sea-going plank boats and ships,which have supplantedthe oldertypes, as has occurredin the Kei Islands. The generalabsence of outriggersfrom the small craftof the harboursand rivers of Borneo is not so easy of explanation. I feel considerablehesitation, however, in making definitestatements concerning the distributionof outriggercanoes in certain areas, as I have been able to findvery little positive evidence one way or the other, and negative evidence is full of pitfalls. Accordingto Modigliani,canoes are veryrare in Nias, and he does not mention an outrigger(1890, p. 418); otherauthors do not appear to mentionthem either. Rosenberg(1878, I, p. 176) figuresa canoe, abak, fromMentawei, with a single outrigger,two booms and a direct lashed attachment. The Mentaweiwar-ship, knabat bogolu,figured by Rosenberg (1888, P1. VIII, Fig. 9), is referredto on pp. 74, 78, 83; it has apparentlya single outriggerwhich is on the starboard side and a direct lashed attachment. Modiglianidescribes the simplebut seaworthycanoes, eloha,of Engano; they are dug-out~swith a double outriggerwhich rests in fourdeep notchesin the gunwale (1894,p. 206). A modelin the Leiden Museumshows two boomsand floatsfar away fromthe hull with a directtied attachment. Giglioli,in describingthe specimens collected by Modigliani,says: " These are long narrow dug-outs [cobara-eloha], madefrom a singletree-trunk, with two outriggers;the cross-piecesor seats are often beautifullycarved, the usual figurebeing that of a man witharms and legs extended as if to keep apart the sides ofthe canoe; grotesquefigures in-coitu or that of a frog being swallowed by a snake are amongstthe specimensof these quaint Engano VOL. L. H

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 98 A. C. HADDON.-The Outriggersof IndonesianCanoes. boat seats. No sails are used, and the paddles are plain and of the usual type" (1893, p. 131). At the present day, outriggersappear to be scarce in Sumatra. Friederici (II, p. 235) states that outriggershave almost disappearedin the Singapore,Banka and Bilitonareas. Dampiernarrates in his Voyagesthat the Nicobarcanoe in which he sailed to Achin had " good outlayerslashed very fast and firmon each side the vessel, being made of strongpoles. So that while these continuedfirm, the vessel could not overset . . . we weretherefore much beholden to our Achinesecompanions forthis contrivance " (quoted fromKloss, pp. 267, 268). This was in 1688; that the " outlayers" were true outriggersand not outlayers(p. 76) seems probable,as in his descriptionof his visit to the Nicobars,Dampier speaks of the " small slight outlayerson one side " of the local canoes (loc. cit., p. 260), and, as we have seen (p. 78), the Nicobarese canoes have single outriggersor none at all. Folkard (p. 480) refersto the long narrowjellore and ballelangof Sumatra which" are fitted with double outriggers,which stand out a considerabledistance fromthe sides." He figuresa jellorewith the sail partlyfurled, the ends ofthe two boomscurve down- wardlyand apparentlyare insertedinto the floats. Folkard does not say wherehe saw these craft,probably it was on the coast of the Palembang district; he adds, " jellores have sometimes only one outrigger." Other sailing vessels without outriggersare the panchallangand the bantang. In the AmsterdamMuseum there is a model of a canoe fromPalembang with a double outrigger,two booms and a directtied attachment. Giglioli (1893, p. 116 and Fig. 8) describesthe Batak solu dug-outswith a wash-strakebut no outrigger;the bow and stern decorationsare noteworthy,they consist of stickswith tufts of hair and a central phallus; the bow in additionhas a wooden carved and painted buffalohead. Brenner(1894,'p. 284) refersto two kinds of keel-lessdug-outs among the Batak of Lake Toba, the solu ratsaranor fishingcanoe, and the solu bolonfor trade and war; some of the latter, he says, are providedwith an outrigger,but he gives neitheran illustrationnor a description. Fischer(p. 114) refersto a modelin the Leiden Museumof a sail boat, djongkang,from Padang, with a double outriggerof two feeble bowed booms; Dr. Juynbollinforms me that these are half insertedand attached by a nail to the float. Hornell's observationthat North-CentralJava is a localitywhere the outrigger patternof canoe has long been discardedin favour of properlybuilt boats (1919, p. 98) supportsFriederici's remark that outriggershave almostdisappeared from the Javan coasts (II, p. 235). Referenceshave been made (pp. 82, 87) to a degenerate canoe fromNorth Java describedby Hornell. There is a distincttype of attachmentin the Eastern Javan area (Madura Strait,Maduxa, and the Bawean Islands) whichhas alreadybeen described(pp. 92-94, Figs. 18-21). Miller remarks(p. 244) that the Bawean and Madura boats are quite isolated in the westernarchipelago, and onlyfind affinities in the extremeeast in the

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 99

South Philippinelocal group. In both areas the foreboom has a directattachment, while the aft boom has an indirectattachment. Mullertakes the Sangir model (Fig. 2, D) as a parenttype from which the EasternJavan and South Philippinetypes have diverged,and appears to regardthe upwardlycurved aft booms of the Javan boats as the equivalent of the upwardly curved " Biigel " (boom-spars,p. 73) of the South Philippinearea (Figs. 2, A, and 7). It seemsto me much morereasonable to regardthem as true booms, or boom-prolongations,which are recurved,and not as reducedboom-spars of the Sangirmodel, of which the outrigger-boomand oblique stays have disappeared. In my opinion a much closer analogy is to be met with in the NorthernCeleban types illustratedby Figs. 22 and 23, the main difference being that in the latterthe attachmentspar is tied to the floatas in the ordinary Halmaheran attachment,whereas in the Eastern Javan type the attachmentrod appears to be always insertedinto the float. The rod-attachmentof the Kema (North Celebes) canoe is associated, as we have seen (p. 96), with a Halmaheran attachment. The Eastern Javan area is contiguousto and partly overlappedby that of the Balinese attachment(p. 88). Juynbolldescribes (p. 37) a model in the Leiden Museumof a tradingvessel (Madura, paduwang fromEast Java) with a double outriggerof two booms whichare tied to the float. In Madura and Bali the attachmentmay be direct and lashed (p. 83). In Madura Strait the straightfore-boom may be lashed to and the curved aft boom insertedinto the float-i.e., a mixed direct attachment(p. 84). Typical of Bali is the Balinese attachment,which consists of a spar rigidlyfastened by one end to the boom, the otherend being insertedinto the float; it also occurs at Lombok (p. 88). A mixeddirect lashed and Balinese attachmentoccurs at Madura (p. 84). A mixeddirect lashed and rod-attachmentoccurs in MaduraStrait and at theBawean Islands, the rod being lashed to or insertedthrough the boom and insertedinto the float (p. 92). We may thus definean Eastern Javanese area which includes the extreme easternend of that island, Madura, Bawean, and Bali, and is characterizedby the occurrenceof attachmentswhich form a gradationof types: (1) In Bali both booms have the sedek. (2) The foreboom is lashed directlyto the float; the aft boom is (a) insertedinto the float,(b) a prolongationof it, the sedek,is insertedinto the float, (c) a rod is insertedinto the floatand insertedinto or lashed to the boom,or (d) the aftboom is lashed directlyto the float. It is alwaysrisky to suggestan evolutionary series,but it looksas if an inserteddirect attachment, here usually under the modified formof the sedek,might have been the earliestform for both booms. The greatest strainin an outriggeris at its foreend, and a lashed attachmentmight have been adopted to counteractthis, while the assumedprimitive form would be retainedfor the aft boom; it will be noted that a lashingis providedfor the sedekfor greater security. The rod attachmentis probably a modificationof the sedek. It would onlyrequire that the rod shouldbe lashed to both the boom and the floatto convert H2

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 100 A. C. HADDoN.-The Outriggersof IndonesianCanoes. it into a simple formof Halmaheran attachment. The lashing would give at the same time greaterstrength and elasticityand probablyless liabilityto fracture. The colonizationof Java fromIndia, accordingto Havell,was probablya sequence of the finalcollapse ofthe Saka powerin India at the beginningof the fifthcentury, whenthe kingdomof Sarashtraor Kathiawar,which had been ruledfor centuries by the Saka dynasty,of foreignorigin, was conqueredby Chandra-guptaII (Vikra- maditya) between A.D. 388 and 401. This great monarch,although tolerant of Buddhismand Jainism,was himselfan orthodoxHindu (V. A. Smith,p. 292). " After that Brahmanismsupplanted Buddhism as the principalState religionof India, the Buddhistart traditionswent with the Saka immigrantsto Java, wherethey reached their highest expressionin the magnificentsculptures of Borobudur" (Havell, p. 113). " ' It havingbeen foretold,'say the [Javanese]chronicles, 'to a kingof Kfij'rat, or Gujerat, that his kingdomwould decay and go to ruin altogether,the Prince resolved to send his son to Java . . . and embarked him with about five thousand followersfor that island. Amongthese followerswere people skilledin agriculture, artificers,men learnedin medicine,able writers,and militarymen. They sailed in six large ships and upwardsof a hundredsmall.' [This was in A.D. 603. Later on a reinforcementwas sent oftwo thousandpeople.] 'From this period,'continue the chronicles,'Java was knownand celebratedas a kingdom; an extensivecommerce was carried on with Gujerat and other countries.. . During the sovereigntyof the Prince and his two immediatesuccessors, the countryadvanced in fame and prosperity. . . artistsespecially in stoneand metalsarrived from distant countries and templeswere constructed . . . ' and at Borobudurin Kedu duringthese periods by artists from India.' . . . The building of the splendid shrine of Borobudur, the most magnificentmonument of Buddhist art in the whole of Asia, is ascribedto circaA.D. 750 to 800, but the decorationof it musthave spreadover several centuries. It was notin factentirely completed before the Buddhistfaith in Java was superseded by orthodoxBrabmanism as the State religion,about the tenthcentury " (Havell, pp. 111, 112). C. Leemans,however, states that, accordingto the annals of Java, Brawidjaja of Kalinga foundedthe Empire of MendangKamoulan in Java. This princearrived in the year 525 of Saka (A.D. 603). What transferredthe religionof Hindustan to Java was not war, these were not conquests,it was commerceand navigation(p. 541). The fall of the Empire of Borobudurtook place towardsthe end of the tenth century(p. 537). Assumingthat there was a large organizedexpedition from India to Java in A.D. 603,it presupposesa knowledgeof the island and ofits suitabilityfor colonization, and foran undeterminedtime previouslythere must have been voyagesto and fro. At all events,we can date the sculptuxesof the ships at latestwithin the eighthand tenth centuriesand the types of the ships may have been commonmuch earlier. These carvingsare of especialvalue in the presentconnection as theyare the earliest

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 101 recordsof outriggers,of which several varieties were fittedto these ocean-faring plank-builtships. Representationsof seven ships are givenin Leemans' atlas of " BorO-Boedoer" (someof which have been coqpiedby RadhakumudMookerji), two ofwhich are without outriggers,one havinga simplemast and the othera tripodone. The otherfive ships have outriggerswhich we may suppose were double, as fourshow a port outrigger, and one a starboardoutrigger. Four have two masts, the other(26) has a single one; one (24) clearlyhas a double or bipedal mast,and in othersit is possiblethat theremay be bipedal or tripodmasts, but the details are obscure. In two a single series of rungsproject fromthe single or both masts, and also in the aft mast of another(27), but in thisship the foremasthas two rungs. In one ship (28) the rungs appear to abut againstrope, it is possiblethey may have been connectedwith it. Three ships have threestraight outrigger-booms, another (27) bas fourbooms, whilethe threebooms of the fifth(28) do not appear to be straight. All thesebooms pass below or over a gunwaleboard. In two ships (26, 27) the straightbooms pass

K

FIG. 24.-INDO-JAVANESE SHIP: TWO BIPED MASTS, THREE STRAIGHT BOOMS WHICH PASS BETWEEN THE TWO LONGITUDINAL sPARS AND OVER THE DOUBLE FLOAT, THREE CURVED BOOMS WHICH PASS OVER THE TWO LONGITUDINAL SPARS AND OVER THE DOIUBLE FLOAT (LEEMANS, pl. Ciii, 176).

below the singlelongitudinal spar, in two (24, 25) theypass betweenthe two longi- tudinalspars. In two ships (25, 27) the ends of thesebooms lie well above the float, in one (26) they appear to pass behindthe innerelement of the double float,and in a fourth(24) theyappear to pass in frontof both elements. In four ships,in associationwith the straightbooms are an equal numberof downwardlycurved booms, which usually pass over the gunwaleboard and under

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 102 A. C. HADDON.-The Outriggersof Indonesian Canoes.

FIG. 25.-INDO-JAVANESE SHIP: TWO MASTS, THREE STRAIGHT BOOMS WHICH PASS BETWEEN THE TWO LONGITUDINAL SPARS, THREE CURVED BOOMS WHICH APPEAR TO PASS BETWEEN THE LONGITIUDINAL SPARS, THEY PASS BETWEEN THE TWO ELEMENTS OF THE DOUBLE FLOAT AND C URL UP TOWARDS THE 'UPPER ASPECT OF THE OIUTER ELEMENT OF THE FLOAT (LEEMANS, pl. Ci, 172).

FIG. 26.-INDO-JAVANESE SHIP: ONE MAST WITH RUNGS, THREE STRAIGHT BOOMS WHICH PASS BELOW THE LONGITUDINAL SPAR AND APPARENTLY BELOW THE DOIUBLE FLOAT, TWO FALSE BOOMS, THREE CURVED BOOMS WHICH PASS OVER THE LONGITUDINAL SPAR AND BETWEEN THE TWO ELEMENTS OF THE FLOAT (LEEMANS, pl. ccli, 41).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 103

<~~~~~~~~~

FIG. 27.-INDO-JAVANESE SHIP: TWO MASTS WITH RUNGS, FOUR STRAIGHT BOOMS WHICH PASS BELOW THE LONGITUDINAL SPAR, FOUR CURVED BOOMS WHICH PASS OVER THE LONGITUDINAL SPAR AND BETWEEN THE TWO ELEMENTS OF THE DOUBLE FLOAT AND PROJECT BEYOND THE OUTER OF THESE (LEEMANS, pI. Cxxiii, 216).

FIG. 28.-INDO-JAVANESE SHIP: TWO MASTS WITH RUNGS, THREE BOOMS WHICH ARE PROBABLY SLIGHTLY CURVED, THEIR ENDS ARE PRESUMABLY LASHED TO THE DOUBLE FLOAT (LEEMANS, p1. lxviii, 106). anotherboard or rail. Whenone longitudinalspar is present(26, 27) theycurve over (in frontof) it, or if there are two longitudinalspars they curve over both in one and doubtfullyover the innerand behindthe outer (i.e., they pass betweenthem) in the other. In two cases (25, 27) the curvedbooms pass betweenthe two elements of the double floatand curlunder and moreor less to the frontof the outerelement;

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 104 A. C. HADDON.-The Outriggersof Indonesian Canoes. in another(26) theysimply pass betweenthe two elementsof the float; in the fourth (24) theypass in frontof both elements. In the fifthship (28) the longitudinalspar is absent and the ends of the slightly curvedthree booms pass to the outer side of the double float,the two elementsof whichare shownas beingtied together-I regardthese three booms as the equivalents of the curvedbooms, the straightbooms being absent. In one (26) of the two ships witha singleinternal longitudinal spar, this spar is furthersupported by two falsebooms to whichit is tied, and a centrallashing passes between this spar and the float, embracingboth elementsof the latter. These contrivancesfrequently persist in Indonesiancanoes. It is verydifficult to arriveat a satisfactoryconclusion as to the constructionof these vessels and to the real nature and structureof the outrigger. One gets the impressionthat the sculptorsdid not reallyunderstand the details of the working drawings,which evidently were suppliedto them. Probablythey were artistswho had no practicalknowledge of sea-craft; further,there were technical difficulties to be overcomein representingso complexan object as a two-mastedsailing ship with its gear in relativelylow relief,and it is amazingthey did it so well. It is evident that the straightbooms, curved booms, and longitudinalspars must have been lashed together,and the ends of the curvedbooms to the float,but theseusually are not indicated. The straightbooms in two ships resemblethe booms of existingboats witha Halmaheranattachment, in two othersthey appear to reach the double float,but whetherthey were actually lashed to it is uncertain. The mostproblematical of these is Fig. 24, here the curvedbooms apparentlymeet the straightbooms on the float; it bears a superficialresemblance to Rosenberg'smodel of a sailingcanoe fromthe MentaweiIslands (Fig. 1), the double floatof whichis apparentlylashed to one boom comingout horizontallyfrom the gunwaleof the canoe and to anotherslanting down fromthe roofof the cabin; it is possiblethat this slantingboom or " boom spar " maycorrespond with the curvedboom ofthe carving;the modellacks the longitudinal sparswhich I regardas beingpresent in the Indo-Javanvessel. What is the natureof the curvedbooms ? At firstsight they appear to be fairly typicalHalmaheran attachments, but, on the otherhand, in threecases theyseem to come out of the side of the vessel,but this is not so markedlythe case in the fourth case (Fig. 24). If theycome out ofthe vesselthey may be regardedas trueoutrigger- booms,somewhat analogous to the centralbooms of the Sulu canoes,or as the curved booms of Pritchett's" pirate craft" (Fig. 8), but if this be so the straightbooms have to be accountedfor. If the curved booms are the essentialbooms, then the straightones may have been intended simply to strengthenthe outrigger. We must rememberthat these craftwere ocean-goingsailing vessels,and consequently the outriggerhad to be verystrong. Owingto the kindnessof L. W. Jenkins,of the Peabody Museum,Salem, Mass.,

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDoN.-The Outriggersof IndonesianCanoes. 105

I am able to illustrate(Fig. 29) a canoe model fromNonuti (Nanouti) Island, GilbertGroup, which has a singleoutrigger of three curved booms, which are attached to the floatin the same manneras that of the Funafuticanoe (Fig. 33). Thereare in additiontwo straightbooms whichare lashed on to the shortlongitudinal spar.

FIG. 29.-SKETCH FROM A PHOTOGRAPH OF A MODEL OF A CANOE WITH THREE CURVED BOOMS WITH A DIRECT LASHED ATTACHMENT AND TWO STRAIGHT BOOMS, NONUTI (NANOIJTI) GILBERT GROUP (SALEM).

The otherspars on the curvedbooms forma platform. There is also a " weather platform." This model supportsthe suggestionthat the straightbooms of the Indo-Javanesecraft merely supported the curvedbooms, which appear to have been the essentialones. In the MarshallGroup (pp. 81, 126) the two straightbooms are supportedby insertedsticks. I have already suggestedhow a ilalmaheran attachmentmight arise froman insertedrod-attachment (p. 99). Grantinga pre-existingHalmaheran attachment, whenthe size of the vessel was increased,its upperpart of the attachmentmay have been prolongedso as to enable it to be securedfirmly to the hull of the vessel, as appears to be the case in some of these carvings. One other alternativepresents itself:it is possiblethat the Halmaheranattachment may be derivedfrom a curved outrigger-boom.With the increasein the size ofthe vesseland the assumedaddition of a straightboom to strengthenthe outriggerwe get the apparentlytypical Indo- Javanese arrangement. If lighter vessels were built the straightboom might persistand only the curvedterminal portion of the curved boom be retained; it would then become the means of connectingthe straightboom with the float. Whiencedid these sea-faringvessels which bore Indian immigrantsto Java obtaintheir outriggers ? At the presenttime the SouthIndian outriggeris composed solely of tvo booms with a directtied attachment. The earliestevidence on this

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 106 A. C. HADDON.-The Outriggersof Indonesian Canoes. point I have been able to discoveris the drawinggiven by Lintscotus(II Pars, XCIX, Pl. XIV, D 3) of a boat fromGoa and Cochinwith a singleoutrigger of two booms whichrest on the gunwalesand on the flat upper surfaceof the large float. The methodof fasteningis not clear; it may be by means of staples,but moreprobably by sinnetlashings, in whichcase thoseof the floatwould pass throughundercut holes. This negativeevidence is by no means conclusive. If the bent booms of the Javan ships be true booms then they mightbe comparedwith the modernSouth Indian outriggers,except fortheir number and the curledlower ends supportingthe float. The numberpresents no difficultyas the larger size of the Javan vessels would necessitatethis. So far as I am aware, the arrangementby which the float is supportedby the ends of the curvedbooms is unique,though the threecentral booms of a modelfrom Gowa, South Celebes,are lashedto the undersideof the float (p. 84). If the curvedbooms be an enlargedHalmaheran attachment, it followseither (1) that this methodwas inventedin India, or (2) that it originatedin Indonesia and was adopted by Hindus forvessels sailing-in this region. As previouslynoted, the greatexpedition of A.D. 603 surelyindicates that the countryto whichthey voyaged was well knownto the leaders,probably through centuries of trading intercourse, and if the outriggerwas alreadyestablished in Indonesia it would not be surprisingthat the local Hindus adopted so practicala device to rendertheir large craftmore sea- worthy. An additionalargument for Indonesian origin is suppliedby the presenceof the longitudinalspar, whichwas sometimesconnected with the floatby a lashing. In 1596, accordingto L. (I, p. 35), the Javan craftconsisted of two-masted merchantships, lonco; war vessels: small,with or withouta sail, parao, and large sailingcathar; and small fishingcanoes witha double outriggerand direct,probably tied, attachment,while they carrieda large sail whichgave themmarvellous speed.

The Lesser Sunda Islands. I have been able to gatherbut verylittle information as regardsthese islands. Bali has alreadybeen dealt with. Hornellstates (p. 99) that a Halmaheranattach- mentis commonin Lombok, the only instanceI know of recordedfrom the whole group,but ProfessorS. J. Hicksonhas givenme a photographof Coupang,Timor, in whichit occurs. A directlashed attachmentoccurs in Sumba and Lomblen,and the Moluccanat Wetta. At Baba (Babber) there are the direct lashed, Halmaheran? and, judgingfrom an indistinctphotograph by PfluigEr(p. 146), somethingthat looks like an insertedstick attachment; I cannot make out the details, but if it be so, it is, so far as I know,a unique recordfor this part of Indonesia,unless the vertical spar of Riedel's model fromTenimber is insertedinto as well as lashed to the float. but a true Halmaheranattachment does occurthere (Fig. 4). The Ket and Aru Islands. A direct lashed attachmenthas been seen in a freshwater creek at Totoat, Kei Islands, and a Moluccanat Batu merah,in thesouth-east. A typicalHalmaheran

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 107 attachment occurs at Dobbo (Pulo Wamar), Axu Islands. Doubtless the great developmentof plank-builtboats in this region has had an inbibitingeffect on outriggercraft, as has occurredin North-westIndonesia. From an ethnologicalpoint of view the Kei Islands were originallyPapuan, as theystill are raciallyto some extent,but theyhave been so influencedby cultures comingin fromthe Archipelagothat they may now appropriatelybe consideredas essentiallyIndonesian. The Kei Islanders are noted throughoutthe Archipelago for theirskill in shipbuilding. Wallace gives a descriptionof the constructionof boats by the Kei Islanders and statesthat theyare the best boat-buildersin Indonesia (II, pp. 183-186,see also pp. 92, 159). Langen also gives an accountof their boat-building and says that they supplyboats to the natives of the Aru and Banda Islands (p. 43). Van Hoevell thinksthat the dialectsof the Aru Islands appear to be closelyakin to bahasa tanah. Friedericidoes not findit so, as they,and the dialects of the Kei Islands, have littleto do withthose dialects,like bahasa tanah,which have a direct culture-historicalconnection with Melanesia. But, on the other hand, language, ethnography,and mythologyshow that the natives of the Aru and Kei Islands have a great commonsubstratum with the Alfursof the Moluccas and Minahassa. An importantelement in the Kei Islands was formedby fugitivesfrom Banda, as to a certainextent is provedby the characteristicsand languageof the folkat Eli, Ellat and Fehr in Great Kei (Fr., III, p. 9). The little westernisland of Dobbo has an annual marketwhich is frequentedby tradersfrom all parts; it formsan economic link betweenNew Guinea and Indonesia. The natives are undoubtedlyPapuans, but settlementsfrom Indonesia have modifiedsome of the coastal groups.

The Moluccas.' There are, in the main, threetypes of craft in Banda and Amboina: (1) the plank boats, (2) the outriggercanoe, and (3) the dug-out. (1) The beautifullarge plank boats, 0re-mba7,have a stem-post,which, as seen fromthe side, is broaderand lower than the stern-post. The orembaiof Amboina and Banda have been modifiedby European influence,but in generalreveal the Indonesianoriginal type, this latter persists among the fishingpopulation and in form and constructionresembles the mon of Buka in the Solomon Islands. Friederici givesfurther details of theirconstruction wbich do not concernus here. (2) Outrigger-canoesconsist of a dug-out with moderatelypointed, almost similarstem- and stern-posts,sometimes with additions. In Banda the sides are oftenheightened by planks,but in Amboinawith a bamboo gunwale. Some Banda canoeshave the highbeak ofthe orembai. All have two floatsat the end oftwo booms athwartthe hull,the floatand boom are attachedby pieces of ratan of the thickness of the thumband lashed withspliced liana or thinratan. The Moluccanattachment 1 The followingaccount is largelytaken from Friederici, II, pp. 235-243,and III, pp. 159-162.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 108 A. C. HADDON.-The Outriggersof Indonesian Canoes. is describedon p. 89. The floatsat Banda consist of bamboo,at Amboinaof one or two lengthsof gabbagabba (the mid-ribof the leaf of the sago palm), side by side. Both kinds are bent up in frontlike sledge-runners.The aft boom is usually quite near to the aft end of the float,the foreboom is furtherfrom its foreend. The booms,which are of light wood, are laid over the edge of the hull, sometimesfor strengththey are Iashed with arenga-stringto a pole placed beneath them and betweenthe sides of the canoe. These poles and the booms are tied with lashing ofratan or arengato ledgesprojecting from the keel and side planksthrough the eyes in whichthe lashingis passed; thereis no special name in the Moluccas forthese holedledges, so Friedericiadopts the Melanesianterm of patnati for them. Sometimes Y -shapedwooden forks are lashedto thebooms to carryfish-spears, paddles, and other gear. Originally,and stillamong the primitivefishers of Banda and Ambon,all the parts are sewn and tied together. The commonestnames for outrigger-canoein Banda, Amboina and the neighbourhoodare pra'u (prahui)and hakka,but in the country-speechof Amboina and the Uliassers are also foundtala, talalo, talo, alal, sapu and sapou. (3) In Banda and Amboinathere are also veryrude, in part trough-like,dug- outs called k6lek6le,which are.also called praJuwhen a clumsydouble outriggeris added. Kolekoleis the same wordas kofrakra (p: 117). The Moluccankorra-korra is a vessel ofthe same constructionas a large orembai, but witha double outrigger. The firstand best descriptionand illustrationsof these was givenby Forrest(pp. 23, 67, 83, pls. 4, 5, 10, 12, 15). These craftare especially knownin the historyof the Moluccas and of WesternNew Guinea as the constituent part of the notoriousHongi fleets. In Banda and Amboina a tripod-mast,with a quadrangularsail set obliquely, is in generaluse. The baler is the concaveportion of a largesea-shell, or made ofthe sheathof a palm-leafwhich has the formof the commonPolynesian baler. Friederici (II, p. 238) states that the technicalnautical termscollected by him at Banda are nearly all Malay or Malayized. Correspondingterms in the language of the old inhabitantsof Banda perhapssurvive in a singlevillage in the Kei Islands inhabited by descendantsof the Banda folk. The Amboina termsare also much Malayized. The old true Alfuranterms occur only in the remoteand slightlyaffected villages of Amboinaand the neighbouringislands. Martindescribes an orembaiseen by him in Piru Bay, West Ceram(1894, p. 86), and refersto threekinds of craftat Buru: the sarua is used in Kajeli Bay in the east (p. 258); the fakatorais the boat of the Galelaese who live with the Sulanese in differentkampongs on the south coast west of Tifu (p. 356), he describesthe rig and methodof sailing (p. 358); the prau as seen on Lake Wakollo,Buru, is unusual, the double outriggerconsists of three slightly curved booms, instead of the customary two, with a directtied attachment(p. 329). A directlashed attachmentoccurs on Lake Wakollo in Buru.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 109

The U -Moluccanattachment occurs here and thereon Buru ; as the predominant or exclusiveform of attachmenton Banda, Amboinaand the Uliassers; on Ceram, withthe exceptionsof mostplaces on thenorth coast; Obi. Friedericispeaks of it as sporadicallydiffused throughout the NorthernMoluccas and as beingfound wherever the bahasa tanah is spoken (III, p. 161). On Amboina the attachmentis almost alwaysthle 0-type, whileat Banda the 6-type predominates. At Amboinathe float usuallyconsists of one or usuallytwo lengthsof gabbagabba (miarib of sago palm-leaf). The simpleMoluccan attachment is confinedto Indonesia,but a crosseddouble variety occurs sporadicallyin the West Pacific (p. 129), which Friedericibelieves was carriedthither by his Alfurenmigration. Some formof Halmaheran attachment is knownfrom Banda, Buru, on Amboina, the northcoast of Ceram,Misol, and at Obi it is morein evidencethan the Moluccan attachment. Four booms may occurat Misol (G., No. 267); I do notknow whether thisis generallythe case, but the type ofoutrigger is preciselylike that characteristic of the north-westarea of New Guinea. Guillemard(1894, p. 427) states that the fauna is Papuan and that the inhabitantsof the interiorare true Papuans, but on the coast are a mixedMalayo-Papuan race. FollowingFriederici, we may adopt the termAlfur as the historicalname for the aborigines,or at all eventsfor early inhabitants of the Moluccas. He entersinto a lengthydiscussion (III, pp. 1-4) of the modernabuses of that word,a termwhich merelyindicates the ruderinland hill-peopleas contrastedwith the more advanced coast-dwellers,a distinctionmet with all over Eastern Indonesia,New Guinea,and the largerislands of the South Seas. He also points out that, " these Alfursof the Moluccas and North-eastCelebes are not somaticallyuniform and their languages belongrather to a linguisticfamily. Ethnologicallythey form a fairlyuniform layer of an older evolutionaryor colonizationperiod than the layer of the coast people; and also historicallythey form a positiveunity, as in a large measurethey appear to have contributedto the Melanesian population of New Guinea and the islands furthereast " (III, p. 3). He adds: " The Alfursof Ceram,Ambon, the Uliassers, and Buru, only persistingnow in the interiorof Ceram and Buru, are physically distinctfrom the Malay and Malayizedpopulation of Mongoloid affinity now occupving the greatestpart of the East Indian Archipelagoand the Philippines. They are darkerthan the latter,taller, more powerful,with fineyet stronglimbs and joints, and withoutthe flatnoses of the Malays, Tagals and others" (III, p. 150). Martin says, " I have no doubt that the highlandersof Buru and Seran[Ceram] are closer to Melanesiansthan to Malaysians" (1894, cf. pp. 79, 119, 288). Some still live on the coasts in variousplaces as von Rosenberghas foundin Ceram(1878, II, p. 26), thoughthe majorityas theirname impliesare " bushmen." In Buru and Amblauthe populationhas undergonea considerablemixture since the arrivalof the Europeans. The coast populationof Buru, especiallyat the Kajeli end, is stronglymixed. Amblau has received immigrantsfrom Nusa Laut and

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 110 A. C. HADDON.-The Outriggersof Indonesian Canoes.

WesternCeram. Everywherehere the trade languageis Molucca-Malay,or at least it is understood,but the Alfursof the interiorhave largelypreserved their speech (Fr., III, p. 9). Friederici(III, p. 7) states that the so-calledAmbonese language, bahasa tanah, is split into very numerousdialects on differentislands. The Alfursof Ceram comprisethe Patasiwa in the west and the Patalima in the east. This condition formerlyobtained in the Uliassersand in Amboina.1 Language and traditionshow that the Alfursof WesternCeram and the originalinhabitants of Amboinaand the Uliasserswere one tribe,and till the arrivalof the Portugueseformed an ethnological and somatologicalwhole, with only dialectic differences.The dialect of Eastern Ceram reachesalso to Goram,thus the bahasa tanah of Amboina,the Uliassers,and of WesternCeram, is manifestlysomewhat more remote. Of the dialectsof Western Ceram that of the old Huamual (Little Ceram) appears to have been the mother- tongueof several others,including that of Amboina; it was the classical,a kind of sacred language (as is the kawi, or AncientJavanese, among the Javanese). At the time whenthe Dutch firstarrived, Huamual was one greatgarden, " a Paradise on earth," with 11,000inhabitants and 2,000 warriors,now it is an awfulwilderness in whichno human beingslive (Riedel, pp. 92, 93; Fr., III, p. 8). Fugitivesfrom Banda have mixedwith the inhabitantsof Eastern Ceram,Ceram Laut, and Goram. No trace remains of the originalinhabitants of the Obi (Ombi) Islands (Fr., III, p. 9). Halmaheraand NeighbouringIslands. A directlashed attachmentoccurs at Batjan, Ternate. and on Lake Galela in North Halmahera. The U -Moluccanattachment predominates at Batj an, witha few cases of the Halmaheranattachment; the floatis almostalways of bamboo as in Banda, some- times a second is tied beside the first. At Ternate the U -Moluccanattachment greatlypredominates (Fr., II, p. 239). The Halmaheranattachment occurs at Tidor. In 1599 (Nicolas,II., Pls. 14, 15, 17) therewere at Ternatetwo-masted merchant ships with an outriggerof three booms and a Y-board attachment; war-vessels, carcolle,a double outrigger,the largesthad a fighting-platform,three booms with the usual Y-board attachmentand longitudinalplanks forpaddlers, though others had two booms usually with Y -boards,over which was a longitudinalplank for paddlers,or a direct(?'tied) attachment,in whichcase the paddlerssat on the float; a " gondola," cymbe,with outrigger,two booms and direct attachment; fishing canoes with outrigger,two booms and direct attachment. As I have previously stated,too muchreliance should not be placed on these earlyengravings, but I think it is safeto assumethat the outriggerwas doublein all vessels; we knowit was in some, 1 For furtherinformation on this dual divisionand fora suggestionas to its origin,see W. J. Perry,p. 46.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. 0. HADDON.-The Outriggersof Indonesian Canoes. 111 and probablyin all, ofthe carcolletype. Presumablythe directattachment was tied. None ofthe old books I have consultedindicate a Moluccanattachment, but it would not be safeto base any definiteargument on thisnegative evidence; Weule (P1. 112, Fig. 7) figuresa " Boot von den Molukken" witha double outriggerof threebooms, beneath the middle of which is a longitudinalboard for paddlers; there is a Moluccanattachment, but he does not give his authority,nor the date of the craft. Friedericinoted that the Halmaheranattachment greatly predominated in the northof Halmahera at Tobelo and Galela,though occasionally the U -Moluccanattach- mentoccurs. In Halmaherathe smallledges on the innerside ofthe hull ofthe canoe are not eyed,but cross-barsare placed beneaththem athwart the hull,and the booms are lashed to the cross-barsby arenga or by coconut-fibrestring. The two lateral feet of the triangle-masthave theirbases perforated,through these holes is passed a transverserod whichitself passes throughand is supportedby two shortstanchions whichare lashed to a cross-bar; but in Tobelo and Galela the stems of spikes bent at a rightangle are fastenedto cross-bars,and theinwardly projecting spike is inserted into the hole in the feetof the two masts. Forrest(pp. 9, 18) was the firstto draw attentionto the meritsof the verypractical tripod mast. The upper ends of three bamboo poles of this verylight mast are fastenedclose togetherhinge-wise in such a way that onlyone pole projectsbeyond the others,and to this is attachedthe mast rope. Below, the middle foot is the longestand it can be tilted at any angle as occasion demands (Fr., II, p. 240). A shortcentral wash-strake is added to some outriggercanoes, withtwo notcheson its freeborder in whichthe booms rest. The floatsat Tobelo,Galela, and Kau are fashionedout of bamboo poles slightlybent up in front,or of thin woodenrods shaped like the runnersof a sledge. The forkson whichthe mast, sail, and othergear are laid may be made of a natural branch,a carved piece of wood, or of a branchedstick insertedinto the hollow of a piece of bamboo; these are stronglylashed to tle booms. Paddles, as everywhereelse in the Moluccas,have a crutchgrip. Balers at Halmahera and Ternateconsist of a bucket-shapedlittle basket made out of a Licuala leaf or are cut out of a piece of bamboo. Wooden anchors weightedwith stones are used universallyin Tobelo and Galela, as ironhas not come intouse there(loc. cit.,p. 241). As a matterof fact, similaranchors are widelyemployed in Indonesia. In Weda Bay, Patani, and Buli, canoes have two booms with a Halmaheran attachment,but in some a false-boomis added; in this case the innerlongitudinal spar (whichlies beneath the knee-bendof the attachments)merely extends from one outriggerboom to the other,whereas the outerlongitudinal spar projectsmuch furtherbeyond the booms,and its ends are boundtightly with arenga or ratanlashing to the ends of the float,while a thirdlashing passes fromthe crossingof the false- boom and outer longitudinalspar to the middle of the float. In Weda Bay and Patani the floatsare usuallyof bamboo, bent upwards rather high, in Buli the sledge- ruinnertype is prevalent. Friedericidid not noticethe forkedmast and gear holders

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 112 A. C. HADDON.-The Outriggersof Indonesian Canoes. in Weda Bay and Patani, which,however, occur at Buli. Neitherdid he see a plank (or wash-strake)added to the hull, as in the northof Halmahera,but, sporadically, the edges of the dug-outswere heightenedamidships by the addition of strips of gabbagabba. The attap cabin in the middleof the crafthad closed sides, these are left open furthereast. The tripod-mastis present. The triton-shelltrumpet is used. A simple,often rough, dug-out without outriggers occurs (loc. cit.,p. 242). A shorter,more obtuse formof Halmaheranattachment is found at Ake-Selaka (Fig. 16 C), and here and therethe U -Moluccanattachment. The floatsare mostly ofthe sledge-runnertvpe, but the bambooform with a slightlyupward curve in front also occurs. When mast and gear forksoccur, they are mostlycrescentic. The rectangularsail, as usual, is set obliquely. Everythingelse is as in otherplaces in Halmahera (loc. cit.,p. 243). Friedericipoints out that the peoples of Halmahera and neighbouringislands forma separatelinguistic group, remote from the Malayo-Polynesianlanguages and apparentlyunconnected with any hithertoknown Papuan languages. Ethnologically the Alfursof Halmabera forma great group with the Alfursof the Moluccas and those of North-eastCelebes, and in \themain probably anthropologicallyas well. He couldnot see thatthe natives of Tidor, Ternate, and Halmaheradiflered appreciably in appearancefrom other Indonesians, and in none of the places that he visiteddid he note featuresor skin-colourapproaching those of the so-called " Aryan race" (III, pp. 9, 10). " There are at least five dialects among the tlfurs of Minahassa. The more westerlyand soutbernlanguages of Celebes (tbose of Gorontalo,Baree, and the Togian Islands) are furtherremoved from Melanesian languages. The dialectsof the Alfursof Minahassa and of the Sangirand Talaut Islands belongto the great group of Philippinelanguages, which, acc6rdingto Kern, reach to Formosa. But the bahasa tanahand the dialects of Minahassa have a commonlinguistic substratum, so that originallythe Alfursof Ceramand its neighbourhoodwere akin to those of Minahassa" (Fr., III, pp. 10, 11).

Celebesand NeighbouringIslands. The aboriginalinhabitants of the Sulla Islands belongethnologically to Middle Celebes; the coastal folkare a verymixed crowd(Fr., III, p. 9). Thereis a Halma- heran attachment. Friedericisaw three kinds of craftin Buton: (1) a boat withoutoutrigger, (2) boats with a double outriggerwith two kinds of attachment(a) with usually threebooms, the U -Moluccanattachment and threebamboos bound together to form the float,(b) an attachmentwhich recalls the Halmaheran(Fig. 16, A). In Makassar he saw no outriggerboats, lepdlep4,but accordingto models in the museumat Weltevredendouble outriggersoccur. In Bugi also the lepalepa'is found(Matthes, Boegi,p. 544, quoted by Fr., II, p. 235).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 113

In Celebesa directlashed attachmenthas been recordedat Makassar and Gowa, a model fromthe latter place with five booms and a mixed direct attachmentis referredto on pp. 80, 84, 106, 114 ; it is foundon the Malili River (at the north-east of the Gulfof Boni, ODM atanna Lake, furtherinland, at Paloppo (on the west side ofthe head ofthe Gulfof Boni and at Libukangin the same gulf,where the Sarasins illustrateone canoe witha floatcomposed of fourbamboos; it also recursat Kema in the extremenorth-east of the island. The Halmaheranattachment, however, in various forms,predominates throughout Celebes. Accordingto the illustrations ofD'Urville (II, pl. XXV, 3) and Pfluiger(pp. 100, 103) the canoes on Lake Tondano, Minahassa,are simpledug-outs with square boardedends. The " praus in roadstead of Makassar" figuredby Pfluger(p. 71) have a triangle mast, the yard may be loweredto the boom or the boom raised to the yard. The Celeban canoes almost invariablyhave two outriggerbooms, the model fromGowa just mentionedbeing very exceptionaland probablyrepresents a Sulu craftand not a local one. The outlayers of sailingboats in the Gulfof Boni are referredto on p. 76. The region embracin,gMinahassa, Banka, the Talaut (includingSangir), and Tulurgroups is characterisedby the occurrenceof a mixeddirect lashed and Halma- heranattachment: thismay, for short, be termedthe " north-easternCeleban area." The examplesknown to me are describedon pp. 95, 96. Muller,as we have seen (p. 98), draws attentionto the affinitiesof the mixed attachmentin this area with that of,the " easternJavanese area "; in both the foreboom is lashed directlyto the float,but in the Javan area the indirectattachment of the aft boomis insertedinto the float,whereas in the Celeban area it is lashed to it, being a typical Halmaheran attachment. Thereis, however,at Kema in Minahassaa mixedrod and Halmaheran attachmentin which the fore boom has a rod attachment which apparently is insertedinto the float,while the aft boom has a typical Halmaheranattachment (Fig. 23). There was much going to and fro in Indonesia in formertimes, so no surpriseneed be felt if these two regions,separated by the whole lengthof the large island of Celebes,show some resemblances,but these are not close in detail. A direct lashed attachmentoccurs at Sangir and a mixed lashed direct and straightvertical Halmaheran attachment. Morethan two booms are present. Minia- turesakit canoes fromNanusa Island (to the north-eastof Celebes)have two or three boomswhich are simplypegged on to the float; the modeof attachment has no signifi- cance,as theseare merelyceremonial models; somesakit canoes are without outriggers, they are employedin the exorcismof the sakit or spirit of sickness(pp. 77, 96, and Hickson, 1893, Fig., p. 290). S5uluArchipelago. In the Sulu archipelagosmall canoes may have but two booms amdthe outrigger maybe placed at the aftend ofthe canoe (Guillemard,1889, p. 206; G., Nos. 180, 217). Usually thereare threeor four booms (p. 80). In these canoes the fore and aft booms are always straight,whereas the centralbooms are downwardlycurved at VOL. L. I

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 114 A. 0. HADDON.-The Outriggersof Indonesian Canoes. theirends (Fig. 7), but in Wilkes' figure(V, p. 333) all threebooms are downwardly curved-this is probablyincorrect. In all cases thereis a directlashed attachment. Two or more upwardlycurved bars are generallyfastened transversely across the canoe, usually they are attached to the booms (when they may be called " boom- spars," they are the " Bigel " of Muller) and frequentlythey bear, near each end of the boom,a crutchor forked stick to supportgear; in thecanoe figuredby Wilkes each ofthe three booms supports a semicircularspar with forked ends, the arms of which in the sectionalview (V, p. 332) are shownas connectedwith the boom by a lashing. An analogouscombination of straightand partiallycurved booms is foundin the Madura district,where the straightbooms are tied to, and the curved ones inserted into, the float(p. 84). In a model fromSouth Celebes,the fore and aft straight booms are insertedinto the float,while the middlethree are lashed to it (p. 113). It is possible that the latter may throw some light on the somewhatperplexing differencein formof the Sulu booms,and one is temptedto suggestthat the outer straightSulu booms were originallyinserted into the float,but this would be the exact oppositeto the arrangementat Madura. Probably belongingto this region is the " pirate craft" of Pritchett,which has threestrongly curved outrigger-booms on each side, the forkedends of whichare lashed to the floats(pp. 81, 84). Guillemardstates that the southernpart of Sulu Island " is inhabitedchiefly by the Bajaus or sea-gipsies,a people quite distinctfrom the Sulus, and of a much lowertype " (1894, II, p. 90).

Philippines. Muller(p. 244) says that the canoes of Sulu, Zamboanga,and Cebu are in the main similar. The floats are not attached to the " Biigel " (Fig. 2, A-C) but are attacheddirectly to the booms[by a lashed attachment],which are not morethan two in numberand are unusuallythin and fragile,and need strengtheningby the elas- ticityof the concave " Bigel " [" boom-spars"] whichare halfthe lengthof the booms. Vojnich figuresthree booms at Manila and on Lake Lanao in Mindanao (pp. 378, 383). Mr. Henry Balfourhas verykindly made drawingsfor me (Figs. 30, 31) of one of two similarmodels of canoes used on the Pasig River,Manila; the specimenswere collectedby the late AdmiralMaclear and presentedby Mrs. Maclear to the Pitt- RiversMuseum, Oxford. They are fittedwith a double outriggerwhich, however, is providedwith but one boom whichis lashed to the float,the otherend of the float is lashed to a thwartwhich projects slightly beyond the sides ofthe hull,an arrange- mentwhich, so faras I am aware,is unique. The modelsare furnishedwith an awning. The outlayersof Philippineboats have been noted on p. 76, and the remark made by Folkardraises the questionas to whetherthe structuresof the Oxfordmodels may not reallybe outlayers. Judgingfrom an illustrationgiven by A. H. Savage Landor, a canoe with a double outrigger,two booms,and a directlashed attachmentoccurs in the Cagayan Group,Mindoro Sea (I, p. 228).

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 115

The most widelyspread, important, and best knownlanguages of the southern Philippinesis that of the so-called " pirate tribes" of Mindanao and Sulu. It is a dialector formof Bisaya witha stronginfiltration of Malayan elements, and is spoken by theMoros of Mindanao, Basilan, in the Sulu and Tawi-TawiArchipelagoes, Palawan, Balabec, and in NorthBorneo (Fr., III, p. 11).

Borneo. As outriggercanoes are such a featurein Indonesia,their almost total absence fromBorneo is rathersurprising, but this absence, as in the westernand northern

FIG. 30.-MODEL OF A CANOE WITH A DOUBLE OUTRIGGER AND ONE BOOM, PASIG RIVER, MANILA (SKETCH BY H. BALFOUR, OXFORD).

FIG. 31.-SAME AS FIG. 30 SEEN FROM ABOVE AND WITHOUT THE AWNING (SKETCH BY H. BALFOUR, OXFORD).

portionsof both Sumatraand Java, seemsto be due to secondarycauses. Hose and McDougall give an illustration(copied fromLing Roth, I, p. 144) of a roughmodel made of pith of a small canoe withtwo outriggers,the booms of whichpass through thesolid hull and are insertedlaterally into the floats. The model,which is in the Leiden Museum, came from South-east Borneo, a districtfar removed fromthe Malanau country. Apparentlya similarimitation canoe, jong, is employedby the Malanau in the bayohceremony for the castingout of diseases (II, Fig. 84, p. 133). I 2

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 116 A. C. HADDON.-The Outriggersof Indonesian Canoes.

They state that the Malanau are Klemantans,a mixedgroup of Borneanpeoples, of the coast regionsof Sarawak, most of whom have recentlybecome converted to Islam, wlile all ofthem have been muchinfluenced by Malays (II, p. 129). A. E. Lawrence, of Sarawak,kindly gave me two sketches (Fig. 32), drawn frommemory, of toy " schooners,"jong, sailed by Malanau and Malay childrenalong the Sarawak coast, bothhave a singleoutrigger, one withtwo straightbooms, while the otherhas onlya singleboom; in both cases the booms are insertedinto holes in the side of the float, pelempong. I am also indebtedto Ivor H. Evans, of Perak,for informingme that when travellingin the TempassukDistrict of BritishNorth Borneo, some Bajaus temporarilyfitted a small low dug-out,gobang, for him with a double outriggerof two straightbooms which were lashed to bamboo floats. This was the onlyoccasion

FIG. 32.-TOY CANOES, jong, OF MALANAU AND MALAY CHILDREN, SARAWAK (FROM SKETCHES BY A. E. LAWRENCE). on whichhe had seen an outriggerin Borneo,but he had not verymuch to do with the villagesnear the coast. He adds, " The Bajaus weregreat pirates in the old days and mighthave picked up the idea in any of the islands furthereast." The word gobangis the Sulu guban (infra),thus the canoe and its name may be an importation, or may be indigenousto the area as a whole. The outlayersdescribed by Beeckman have been noted on p. 76. For a descriptionof the usual type of craftin Borneo see Ling Roth (II, pp. 246-254),and forceremonial boats (I, pp. 144, 283); he states, " It would seem at one time Sumatra was supplied with boats fromBorneo " (II, f.n., p. 249); also Hose and McDougall (I, pp. 55, 56, 132, 166), Gomes (pp. 49-51) and Nieuenhuis.

NOTESON THE NATIVENAMES FOR CANOES AND OUTRIGGERS. The followingterms for outrigger canoes are given by Friederici,the names for boats withoutoutriggers are placed within( ) Terms of outriggercanoes added by me are placed within[ ]. Anda&mans [chM-rigma].Nicobars [due, dae and doai]. Mentawei [abak]. Engano [eloha]. Sumatra [jellore, ballelang]. Bali, dyukgn,sdmp"n in Malay. Confinedto a restrictedarea of the Lesser Sunda Islands, we findin Sumba, tena, teneh; Middle and Eastern Flores: Sikka, tena,and Maumeri,tenah; Solor,tenna.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 117

Flores, korakora. The most customarynames for an outriggercanoe in Banda, Amboinaand the Uliassersare prau?(prahA) and hAka,;but also tala, talalo,talo, alal, sapu and sapou are to be foundin the countryspeech of Ambon and the Uliassers (II, 237). Sumbawa,lopie or lopi. Banda, korakora,prau (kolekole,but called prau whenoutrigger added; orembai). Amboina,prau, prah'usema4n, prahu belan,haka, kolekole,korakora, lepa or prahu lepalepa,tala, talalo,alal, sapu (orembai). Batjan, nyYn. Ternate, oti or 6ti. Halmiahera: korakora; Weda Bay and Patani, yel; Buli, peldn; Tobelo, ngot,ri,6tili or otil, (here the lepcilepais a dug-outwithout outrigger); Galela, deru; Ake Selaka, oti. Celebes, lepalepa, korakora; Salaier, lopi. Sulu, guban. The followingnotes are taken mainlyfrom Friederici (II, pp. 244-246),who gives fullreferences. He also gives the variantsof these termsthat occur in Melanesia, but these do not concernus here. The dyukilnof Bali does not occur in any formeastwards of the Philippines. It is the Malay djokong; Javanese djukung,djungkung; [Nicobar,due]; Batjan, nyon; in the Philippines: Tontemboan,rungku, Bisaya and Bicol, adyong; Iloko, daon; Tagal, dawong; [Borneo,jong]; Diad6kqi or diadAk4&is the name for the outrigger-boomsat Tobelo (Halmahera). Also cf. jonco (p. 106). Kolekoleis clearlythe same wordas k6ra'korc,which term is currentin Makassar, throughoutthe Moluccas, in Halmahera and Western New Guinea, and it also extendsto Flores. It is more than doubtful,says Friederici(III, p. 159)-,whether the word korakoraof the Moluccasreally goes back to the Arabic qorq6rand thence to the Portuguesecarraca. Valentijn,who treatsthe word as an indigenousone, is certainlywrong when he derivesit fromthe Malay kura-kura. In the oldest Portu- gueseand Spanishaccounts of the Moluccaswe findcaracora, coracora, carcoa, but never carraca; de Morganot onlysays expresslythat it is an ancientand indigenousword amongthe Tagals ofMindoro, Marinduque, and Luzon,but thatit is also a trueMalayo- Polynesianword: in the Malay Peninsula,kolek (a small fishingboat); Amboina, kolekole; Mota (Banks Islands) kora; San Cristoval(South Solomons),ora. Friede- rici asserts that this equation is not merelyaccidental as there is other support forit, and he has shownthat most of the boat-namesof Eastern Indonesia recurin Melanesia. He adds, (III,* p. 159) that the Moluccan korakoracorresponds to the Philippinebaranguay. [The Achinesehave a sailing boat called " kolay" (Folkard, p. 481)]. The Malay prahu,perahu, p.rahu, prau, etc., is 'a generalterm for canoe, boat, or-shipand appearsto be used indiscriminatelyin places forcraft with or withoutan outrigger,and thereforehas no special significance,thus in Amboina an outrigger canoe may be termedprahit semran. The h4ka-of Amboina is the Malayo-Polynesianwangka, which is so widely spreadin the South.Seas as to be universal. In Saonek and Sorong,wa?,, and Manuk- wari (Doreh) wa or wai: all in New Guinea. [Ray points out that the termlaka,

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 118 A. C. HADDON.-The Outriggersof IndonesianCanoes. used in Madagascarfor the canoe witha doubleoutrigger, " is no doubtthe Melanesian waka,etc., also that Malagasy and Tagal are moreclosely related than eitherof them is to the language of the islandsbetween" (Haddon, 1918, p. 53). Christian(p. 229) gives: Malay wangkang,a junk; Moltuccas,waga, a vessel; PhilippineIslands, banka,a canoe; Waigiu, waag,a piratecraft. Keane points out that the Mentawei abak is the va'a or vaka of Polynesia (1899,f.n., p. 244; 1920f.n., p. 235).] Anotherinteresting word fromthe point of view of migrationsto the Western Pacificis the lepalepa of the Bugi [originallya people ofthe Boni districtof Celebes], it is also employedat Makassar; variantsare: Malay Peninsula,lopeh, lupek, lopek/; at Bima and Sangarin Sumbawa,lopie or lopi; Salaier Islands, lopi; Amboina,lepa, prahu lepalepa; Tobelo (Halmahera), lepdlepa,for a dug-outwithout an outrigger (Fr., II, p. 241); Philippines,lapis, lapes. In Ambon and the Uliassers are alal, tala, talo,talal, talalo; it is possiblethat the wordyel, which is apparentlyconfined to Weda Bay (Halmahera),belongs here. In Halmahera and the neighbouringislands we find: dti or oti, Ternate,Roni, Ake-Selaka; 6ti, Tidor Isam, Tololiku, Waioli; ngo6tili,Madole, Tabaru; n6tili, Ibu; ngMtKr,,6tili and otilTobelo. Apparentlyrestricted to Galela and Loda is the word deru. Obviouslyconnected with po5ldnof Buli in Halmahera are: bero,Solor; belo, berok,prahu berok,Timor; bero,Wetta and Tenimber; prahA belan or belang, Amboina; beri-beriNamatote Island, north-westof Triton Bay, New Guinea; in the Philippines: balanay, Tagal; baranay,Iloko. Kern equates the velo-veloof Fiji [whichbelongs to this series]with Biduq, Malay and biluqTagal (Fr., II, p. 245). Friedericialso groupstogether loju, Sulla Islands; lotu,Gesir, south-east point of Ceram; lete'ie,lettej, Aru Islands. With the termguban of the Sulu Islands may be equated kowa,Sava (between Sumba and Timor); ofa, ofak, Rotti (south of Timor); gobun, Bongu. Later Friederici(III, p. 160,f.n., 280) adds that the gubbanof the Sulu Islands occurs as gobain Malay,and it appearsto himthat the Dayak top,a smallship, is also a related word; [amongthe Bajau of BritishNorth Borneo we findgobang, p. 116]. The beautiful,large plank-boatswithout outriggers of Ambonand Banda come fromthe Kei Islands, but theirname of 0r6mbCaUor 6rermbai, also 0r4nb&ior 4rdnbaik, does not come thencebut is probablyof Malay origin,rermbaja; other variants in the country-speechof Amboinaare: arobail,arubai, arubaillo, arumbai, and as arumbae in Gesir(Fr., II, pp. 235, 236). An orembbaiwith outriggers is called korakora,a word whichis foundas kolekole,kolek in Malayo-Polynesianlanguages (Fr., III, p. 159). Martin(p. 86) says that the name orembaiis a contractionfor orang baik " good friend" (or "servant ") and comesfrom the era ofthe Hongi voyages! The numberof the termsfor booms given by Friedericiis insufficientto enable us to arriveat any conclusions. The Moluccanterm seems to be theitram6n or 6raminn of Banda, whichbecomes bairinan in Batjan and brayu,nanin Bali. Several terms

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 119

are employedin Halmahera,one of whichthe nddyu-nadyuof Ake Selaka extendsas might be expected to Ternate and reaches Amboina, where nadyin also occurs. Concerningthe termse'ri, ari of Buli and sesa of Galela (both in Halmahera)there is nothingto say, but -asFriederici points out, the diddkui of Tobelo seems to be the originof the word kiato,which is so widelydistributed in Melanesia and Polynesia foran outriggerboom. [The Nicobar name is deia due.] Friedericigives the followingnames for outriggerattachments: Bali, sod6k; Banda, Angerg;Amboina, pagupagu; Batjan, tudQtuda; Ternate,pagu; Halma- hera: Buli, t6iW,Tobelo, tY4ntene, Ake Selaka, sek. [The numberof namesavailable is so limitedas to be inconclusive. The sedekof Bali may be connectedwith the seke of Ake Selaka, and Teriiate and Amboina are again linked together. The Nicobarterm is heneme.] As Friedericipoints out (II, pp. 246, 235-243), the termsarman, or some variant of it for the float of an outrigger,is so widespreadamong the Malayo-Polynesian peoplesthat the investigatorshould note whenever it does not occur; it is its variants that need comparison. Thus we have: Banda and Ambon,semln; Batjan, somnan; Ternate, samrn; Halmahera: Weda Bay (and Patani) zom(Yn,somrn ; Tobelo, hamana; Galela, suma; Roni, Ake Selaka, samrtand semasemrain Malay; also variantsas in Sangir, sahermmang. He thinks,contrary to Kern, that the generalterm for outrigger-float is derived fromthe old Javanesesamca, " like," Bali, sama samasama; also Sumba,sama means " at the same time, together,with, at the same time like," etc. For the floatis in realitya miniatureboat, a small counterpartof a boat, indeedit has been stated, froman ethnologicalpoint of view, that the outriggerboat has been derivedfrom the double-boat. So also have we in Malay: sama, " resembling,similar " ; sama- sama, " together,"and in Tagal, sama, " companion,to accompany" and " companion like the servant" ; samaco, " a fitted-upboat " and " to prepareor embellish,to constructla banca (the Philippinecanoe)." The only otherterms Friederici notes forfloat are the bMraof Buli, concerning whichhe has nothingto say, and the kateror katirof Bali, whichis quite obviously: Tara (Baree) katigi; Bisaya, katig; Lake Magindanao,Mindanao, katik; and Bikol, katig. This appears very significantwhen we recall the correspondencebetween certaincraft of Bali and thoseof the area betweenthe Philippinesand NorthCelebes. [The Nicobar name is hentaha.]

GENERAL CONCLUSIONS. It may be taken as a generalprinciple in distributions,whether of animalsor men,that the more primitive,that is to say less advanced, types are generally to be foundon the marginsof an area or in the less accessibleor undesirablelocalities within the area. If the area be an archipelagolike Indonesia we should there- fore expect to find more primitiveconditions on its outskirtsor in the interior

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 120 A. C. HADDoN.-The Outriggersof IndonesianCcanoes.

of the largerislands, and manytravellers tell us that as a rule the latteris indeed the case. As a knowledgeof the ethnographyof Indonesia is necessaryfor a study of that of Oceania, so the converseis equally true. Cultureshave been so crowdedinto Indonesiaat varioustimes, and so manyinternal movements have taken place, that a disentanglingof the chronologicalsequence of the culturesis a very difficult undertaking. A valuable attemptof this kind has been madeby Perry(1918), which, however,has littlebearing on the presentproblem. It is recognisedthat migrations of variable extenttake place everywhere,each always associated with definiteand characteristiccultural elements. We know that similarmigrations from Indonesia have taken place at intervalsto various regionsof the Pacificwhich transported samplesof the then stages of culture. It is froma considerationof such samplesin theseOceanic areas and a co-ordinationof the linkedcultural elements they contain, thatthe relativeage ofethnographical data in Indonesiawill ultimately be elucidated. From differentpoints of view Graebner,Friederici, and Rivers have done great servicein this direction. Graebner(1905, 1909) was, I believe, the firstto utilise the generaltype of canoe, as well as the presenceor absence of outriggersand most obvious featuresof their attachments,as evidencefor the differentiationof various culturesand culture-stratain Oceania, but the details he gives are very meagre. Friederici(II, 1908) has presenteda mass of technologicaland linguisticevidence concerningcanoes, which is ofthe greatestimportance for the studyof these problems, and later (III, 1913) has attemptedto map out certaincultural streams from Indo- nesia into the Pacific. Finally, Rivers (1914) has utilised similar evidencein his dissectionof the culturalstrata in Melanesia. All movementsbetween diflerent islands in Indonesiaand thosethence to Oceania musthave been accomplishedby boats of some descriptionor other,and eventually we may hope to discoverwhat kindsof boats theyemployed, the types of outriggers and varietiesof attachments. My main object in this paper is to preparethe way formore extended historical studies on theselines. The problemsof the distributionand historyof canoes in Oceania are beyond the scope ofthis paper; forthis purpose it willbe necessaryto make a morethorough analysis of the typesof canoes and outriggersthroughout the Pacific,there being many varieties which have to be accountedfor, and, furthermore,a definiteter- minologywill have to be generallyadopted. But formy immediate purpose I findit necessaryto make a fewremarks upon certaindistributions of canoes in Oceania. A verybrief summary of the probableracial historyof the area will perhapsnot be out of place. There seems sufficientevidence to assume that much later than the originaloccupation of a part or the whole of Indonesia by Negritos,Papuans, and the less earlyPre-Dravidians, were the migrationsfrom somewhere in Southern Asia of the dolichocephalicIndonesians. Possibly some of themwere, at all events in their later migrations,already somewhat admixed with southernMongoloids.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 121

Somewhatlater perhaps came swarmsof brachycephalicsouthern Mongoloids, who may convenientlybe termedProto-Malays, some of whom may have been crossed previouslywith other stocks. On thewhole they have dominatedthe true Indonesian peoples; but the bulk of the populationof the archipelagoconsists of variousblends of thesetwo stocks. The Malay Peninsulawas firstoccupied in the twelfthcentury A.D. by the true Malays, Orang Malayu, who crossed over fromMenangkabau in Sumatra; thenceat the close of the thirteenthcentury they spread over the East Indian archipelago. But longpreviously to thisother peoples had secureda foothold in Java and elsewhere. From the firstcentury of our era there were migrations fromIndia. The JavaneseBabads tell of an Indian princewho came to Java about A.D. 78 or 120, wherehe founda nomadicpeople. We know of Indian coloniesin Bali, Sumatra and Java in the thirdcentury (Fritsch, pp. 14, 21). I have already referredto a later colonisation(p. 100). Chineseinfiltration may not have begun after220 B.C., whenSouth China was conqueredfrom the aboriginalpopulation and a seaboard acquired,but commercial relationsexisted with Java and other islands in the fifthcentury A.D., and were continuedfor a long period,perhaps they have neverceased. The ChineseBuddhist missionaryFa-Hien, or Hsien, visitedJava fromIndia in the fifthcentury A.D. Arabiantraders voyaged to the East Indian archipelagolong beforethe time of Mubtammad,but Islam changedthe Arab traderinto a teacherof the new doctrine. At the beginningof the sixteenthcentury the Portuguesemade settlements,and werefollowed later by otherEuropean peoples (Haddon, 1911, p. 35). Two hypothesesare currentconcerning the originof the outrigger: (1) That it is derivedfrom a double canoe, one ofthe canoes havingdegenerated into the float of the outrigger. In double canoes one of themoften is smallerthan the otherand in some places the smallercanoe bears the same name as the float,as for example at Mailu,in BritishNew Guinea,where the termlarima is used forthe smallerelement ofa double canoe, oroft,and forthe floatof a singlecanoe, vaona. (2) That the canoe was evolved fromthe centrallog of a floator raft,the two outermostlogs of which have persistedas the floatsof a double outrigger,an evolutionwhich took place solely in Indonesia. If the firsthypothesis be correct,it would followthat the single outriggerwas the primitivetype, but the second hypothesiswould make the double outriggerthe originalform. But if we assume a dug-outto be the initialboat, and it is or has been used in practicallyevery part of the world(rafts, bark canoes, and skin boats are anotherstory), there does not seem to be any reason why eitherform of outriggershould be the earlier,though in this case there is no structurefrom whichan outriggercould be naturallydeveloped. The question of the priorityof the single or double outriggeris of some im- portance. So far as historicaldata go, the earliestrecord we have is that of the Indo-Javanesedouble outriggerships of twelvecenturies ago. Apart fromthe East Africanarea, to which I have already alluded (p. 78), double outriggersoutside of

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 122 A. C. HADDON.-The Outriggersof Indonesian Canoes.

Indonesia are foundwith a .rareand scattereddistribution in Oceania. They occur in the area whichincludes Torres Straits and the estuaryof the Fly River,but single outriggersalso now occur there, both being associated with an overcrossedstick attachment(p. 85). They are also to be foundin NorthQueensland, (a) the Batavia River type with an attachmentof two vertical sticks,from Batavia River in the Gulfof Carpentaria,round Cape York to about Cape Grenville(lat. 120 S.); (b) the Claremonttype with a direct lashed attachment,apparently from Cape Direction (lat. 130 S.) to ClaremontPoint (lat. 140 S.) (Haddon, 1913). They have been reportedin Oceania, formerlyin the Pelew Islands, and doubtfullyin the Seniavina Group of the Carolines; they occur at Nissan with a direct lashed attachment, temporarilyin the Solomonswith three booms lashed to a floatconsisting of a bundle of bamboos,doubtfully at Samoa, and formerlyat the Marquesas,and also formerly at Easter Island with a direct lashed, attachment(Haddon, 1913, p. 621). A scatteredmarginal distribution of this kind suggestsantiquity, and provisionallywe may accept this supposition. Althoughthe TorresStraits area is not geographically remotefrom Indonesia, I am stronglyof opinion that it is culturallyremote and that the double outriggercame thereby the West Pacificroute. In TorresStraits and the estuaryof the Fly we have definiteevidence that the single outriggeris oustingthe double, and it seems probablethat this has occurred elsewhere,though there is very little precise informationon the subject. Muller (1912, p. 245) alludes to a changeof double outriggerinto a singleas havingoccurred in the Pelew Islands. In a letterto me dated July 14, 1913, Friedericisays: " I have now no doubt that the k6p [the Nissan double outriggercanoe witha direct tied attachment]has been broughtby a Philippine or sub-Philippinewandering stream to New Ireland and neighbourhood,and that the double outriggerhas in courseof time been displaced by the Melanesian single outriggerand has stood its groundonly in the island of Nissan." Apparentlyboth formsoccurred at Easter Island (p. 124). The double outriggeris morestable but clumsierthan the single,the latterhas bettersailing qualities,but greatcare has to be exercisedin the managementof the floatin orderto maintainthe balance of the boat. The single outriggeris a marginalphenomenon. WVithinIndonesia it occurs very rarely and sporadically (p. 78). It is normialin the Andamans and Nicobars and reappears in Geelvink Bay, it is practicallyuniversal in New Guinea and Oceania. Very frequentlyassociated with a single outriggeris the presenceof several booms as in the Andamans,Geelvink Bay, the Massim,West Papuo-Melanesian,and Gulfdistricts of BritishNew Guinea,and mostparts of Melanesia,but to a muchless extent in Polynesia. On the otherhand, we have seen (p. 80) that the Philippine area, includingthe Sulu Islands, SangirIslands (occasionally),in Misol,Waigiu and otherneighbouring islands, and GeelvinkBay numerousbooms are associatedwith a

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-TIheOutriggers of IndonestanCanoes. 123 double outrigger. Thus so far as Indonesia is concernedseveral booms to a single outriggeris a marginalcharacteristic, and so is, but to a less extent,the association of severalbooms withdouble outriggers,as thisis practicallyconfined to the eastern marginof Indonesia. We mustnot, however,overlook the occasional occurrenceof threebooms withinthe centralarea of Indonesia (p. 80). At presentI leave it an open questionwhether these are relics of a more generaldistribution, as I suspect themto be, or as borrowingsfrom the easternmargin. Indonesia, excludingits eastern margin,is thereforecharacterised by canoes having double outriggerswith two booms. The diflusionof canoes with double outriggersand two booms tromIndonesia into Oceania musthave takenplace in veryearly times and possiblyon two occasions, one witha directtied attachment(p. 124) and the otherwith an insertedstick attach- ment. At presentI am not in a positionto suggestwhich is the older type; but if the float be derived from the smaller element of a double canoe an indirect attachmentwould not becomenecessary until the floathad been reducedto a log or a piece ofbamboo. On thehypothesis of the evolutionof the outriggercanoe froma raft, the tied attachmentwould be the moreprimitive, as the need of an indirect attachmentwould not ariseuntil the centralplank of the rafthad become converted into a canoe. On the other hand, a stickattachment appears to be characteristic of varioustypes of primitiveoutrigger canoes. I nowpass to a considerationof the distribution of the main types of attachments.

A.-Direct. 1. Inserted.-Withthe exceptionof the somewhataberrant Balinese attachment I have not come across an unequivocalcase in Indonesia of an actual canoe whereall the booms are so attached. The directinserted attachment has a less extendedrange than the lashed. The only Melanesian record is fromEromanga in the New Hebrides (Hedley, p. 287). Kiramerfound at Tutuila in the Samoan groupthat the straightbooms have a down- wardlyslanting branch the end of whichis insertedinto the float(1903, II, p. 249; 1906,Fig. 42, p. 415; 1911,p. 23a). I have beentold it occursat Rotuma; Friederici also recordsit forMangaia and Rimatara, whilein the morenortherly Cook Islands of Atiu and Mitiarothe end of the boomsis curveddownwardly to be insertedin the float(Fr., II, p. 314, Figs. 127-130). The directinserted attachment also occursin the Paumotus (Pallander, p. 194, figuresa canoe with two booms, the forewith a crooked insertedboom, the aft with a stick attachment). It thus appears to be most prevalentin the SouthernPolynesian area. I feelsome slighthesitation in acceptingall theserecords, as the Tutuila attach- mentis verylike that whichoccurs at Funafuti,which, though it looks as ifit should belongto thisclass, is reallya lashedattachment, and we knowthere have been close relationsbetween the Samoans and Ellice Islanders.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 124 A. C. HADDON.-The Outriggersof Indonesian Canoes.

2. Lashed.-This type is verywidely but sparselydistributed throughout Indo- nesia, so much so that little can be deduced therefrom,unless it signifiesthat it is reallyan ancientform. It is interestingto note that it cropsup in variouslakes and rivers,where its appearancemay mean that it is an old type,or simplythat in these calmerwaters it sufficesfor the needsof the fisherfolk. But it is also significantthat this type is very prevalentin the Sulu Islands and in the southernand central Philippines,i.e., in the easternmargin of our area. The directlashed attachment(with a single outrigger)alone occurs in south India and Ceylon,whence it may have spread into Indonesia, or may equally well have traversedin the reversedirection. It veryrarely crops up in WesternOceania, where it is associatedwith a double outrigger; e.g.,the kopof Nissan in the NorthSolomons (Krause (Fig. 101); Frie- derici,cf. lHaddon 1913, Fig. 14) and North-eastQueensland from 13?-14? S. lat. (Claremonttype, loc. cit.,Fig. 7, afterRoth). It occurs with a single outriggerat Nonutiin the Gilberts(Fig. 29) and universallyin the Hawaiian Islands, and in the Marquesas (model, Salem Museum; ? Pallander, P1., p. 240); at Easter Island witha double outtigger(Choris, P1. X, Fig. 1, cf.Haddon, 1913, p. 621); but in the " Atlas du Voyage de la Perouse" (London, 1798, pl. 61), a singleoutrigger of two booms is shown,the floatrests upon and is tied to the booms; ? Napuka (Wytoo- kee), Paumotus (Wilkes, I, p. 319, but the sketch is not convincing); Tahiti (Wilkes, II, p. 21; Edge-Partington,Ethnol. Album, I, P1. 29); and Huaheine, the easternmostof the Leeward Group of the Society Islands (Ellis, II, 1831, p. 352). At Funafutiin the Ellice Islands (Fig. 33) thereare threestraight booms

FIG. 33.-ATTACIMENTS OF CANOES, FUNAWUTI, ELLICE ISLANDS (AFTER HEDLEY). whichhave a long oblique branchnear the end, the tip of this branchrests on the floatto whichit is lashed, sometimesit is lashed to shortpegs on both sides of the boom (Hedley,p. 286, P1. XV, Figs. 1, 6, 7; Alexander,p. 796, pl. 35); in a model in the CambridgeMuseum there is only one peg; the pegs are insertedinto the float. De Clercq and Schmeltzdescribe (p. 94) and figure(P1. XXIV, Fig. 5) a canoe fromAnsus, New Guinea, with a direct lashed attachment,but as all the accountswe have fromGeelvink Bay speak solelyof a " spike " attachmentwe may suspect that the model is inaccurate. In 1914 I saw at Buniki, a shortdistance

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 125

up the Dabara arumo,a smallriver on the east side ofthe Bebea River (the northerly mouth of the Bamu) in the WesternDivision of British New Guinea, a solitary example of a canoe with a singleoutrigger of two booms with a directtied attach- ment-it may have been an individualoccurrence of no special significance,as the canoe was a verysmall one-another canoe had the stickattachment characteristic of that district. A mixeddirect inserted and stickattachment.-In all the boats seen by Friederici at Tahiti and the Northand South Tuamotus (Pauri-otus)the aft boom had a direct insertedattachmient, but the foreboom had sticks (Fr., II, p. 314, Figs. 131, 132); Pallander (P1.,p. 194) figuresa canoe at Anaa, Paumotus,in whichthe angular(? fore) boom is insertedinto the floatand the straight(? aft) boom is supportedby two pairs of undercrossedsticks.

B.-Indirect Attachment.

(a) AttachmentInserted into theFloat. 1. Stick.-The onlyplaces in Indonesia wherethis occursare the Andamansand Nicobars,i.e., marginalwithin the area; herethe sticksvary in numberand arrange- ment,they may be vertical,oblique or if in pairs overcrossedor undercrossed. In a copy of an old engraving(Mager, p. 137) two double booms,one above the other(boom and boom spar),are attachedto the floatby two stickson the same side of both booms in a canoe of the Mariana or Ladrone Islands; Anson gives a, most excellentaccount and engraving(reproduced by Ling Roth, p. 123, P1., p. 118, and sketchedby Folkard,Fig. p. 463) of " a flyingproa taken at the Ladrone Islands," it has threebooms (the centralone may or may not be a falseboom) withthe same attachment,there are two longitudinalspars and two oblique stay spars; and Safford(P1. XI, p. 493) gives a figureof a Guam canoe withtwo booms each withone pair of undercrossedsticks and apparentlya verticallashing between them. Kubaryfigures (Pls. LIII, Fig. 20; LIV, Figs. 3, 5) a paopao canoe fromNukuor and a warcanoe fromPonape in the Carolineswith a similarattachment of two oblique sticks,the latterl,as two stay spars,to each of which,where it becomesfree from the hull,is fasteneda curvedspar, apic, whichapparently is insertedinto the float between the normalattachment and its ends; a similarspar on each side of the two straight booms occurredon a modelfrom Kusaie in the Salem Museum (but one ofthese apic has entirelydisappeared, while the otheris representedonly by its distal end which is insertedinto the float; how they werefastened to the hull is not shown in the photograph),but the boomshave a Y-stick attachment(these are thus a mixeddirect insertedand stick-,and directinserted and Y-stick-attachments);another Kusaie model (Salem Museum) has two pairs of undercrossedsticks for each boom and a curvedspar whichis lashedto the boom and comesdownwards and forwardsbeneath

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 126 A. C. HADDON.-The Outriggersof Indonesian Canoes. the crossingsof the sticks; at Wolea are two verticalsticks on the outerside of each boom,or two stickson each side, a " bracingspar " slantsup fromthe sticksto the boom (Macmillan Brown, P1. F, and photograph; Mayer, p. 129). The typical MarshallIslands canoe has two straightbooms with one stick(model Salem Museum), two sticks,one on each side (Kriimer1906, p. 416, Fig. 6) or two on each side (model AmsterdamMuseum); in the two formerthree curved booms whicharise fromthe hull on eitherside ofthe boomsand are lashedto the float(i.e., a mixeddirect lashed and stick attachment)see pp.,81, 105, Kubary (P1. LIV, Fig. 7) indicatesa similar arrangement. Folikard(p. 499), as usual, copied Wilkes' drawing(V., p. 49) of a canoe fromDrummond I. (Tapiteuea), Gilbert or Kingsmill Islands, with three convergingbooms and one oblique stickon each side of each boom; a modelin the CambridgeMuseum fromthis group has threevertical sticks,and another model has one pair of undercrossedsticks to each of the threebooms. A simple stick attachmentcharacterizes the tsine canoe of Nissan, an island betweenNew Ireland and Bougainville(Fr., II, Figs. 95, 98; Haddon, 1913, Fig. 6) and the threenortherly islands of the scatteredchain to the east of the Solomons: Nuguria(Thilenius, 1902, p. 61, P1. III, Fig. 2; Fr., II, Figs.106-108),Tauu (Fr., II, Fig. 103),and Nukumanu(Fr., II, p. 300). Thereis intercoursebetween these islands and the Gilbertsand betweenthe latterand the PolynesianIslands to the south, so it is not surprisingthat a similarform of attachmentis commonto them all. Thus we meet with it in Rotuma,Union or Tokelau group,Samoan group,Friendly Islands (Tongan group), and Nieue. The furthesteast that I know of it is from Nikuhiva (Marquesas), where d'Urville (I, P1. LXI, Fig. 2) illustratesa double canoe withfour curved booms and an attachmentof fourtall vertical sticks. In the Fiji group the sticks consist of two pairs of divergingovercrossed sticks; preciselythis arrangementis typicalof TorresStraits, where it is associatedwith a double outrigger,and a two-stickattachment occurs at Batavia River,North-west Queensland,which is remarkablylike the Nissan tsVne,except that thereis a double outrigger. Thereis considerablevariation in the stickattachments of Northern and Southern Melanesia,which lack of space precludesme fromdescribing. The most common type consistsof undercrossedsticks. New Guinea is also characterizedby various types of stick attachment; as I am preparinga memoiron this subject I need not furtherallude to it, except to say that the undercrossedtype extendsfrom the east of GeelvinkBay to the iassim Districtin the extremesouth-east, whence probably came the Cape Bedfordwangga, which extends along the north-eastcoast of Queens- land fromthe FlindersGroup in the south of PrincessCharlotte Bay (lat. 14' S.) to Cape Grafton(lat. 17? S.) (Haddon, 1913,p. 617). The stick attachmentmay certainlybe regarded as an ancient type. Its presencein the Andamans,New Guinea, and Melanesia at firstsight might suggest that it was primitivelyassociated withthe pygmy,or the taller Oceanic Negroids

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 127 who are usually termed Papuans. Among the Negritos only the Andamanese possess it, and the arrangementof theirstick attachmentpoints to a borrowingof the outriggerfrom the Nicobars. In New Guinea and Melanesia tbe evidence is stronglyagainst it being part of the old Papuan culture. Very fewtrue Papuan peoples possess an outriggerof any kind, and the names forcanoes, and especially forthe float,in New Guinea are in the greatmajority of cases of Austronesianorigin We know that duringlong periods of time many migrationshave spread fromIn- donesia into New Guinea and the WesternPacific, the earlierof which gave riseto that mixed folk whom we term Melanesians,and several migrationsof Melanesian peopleshave passed into New Guinea carryingwith them special types of canoes and outriggersand theirdistinctive names. The generalprevalence of the stick attach- mentnot onlyin New Guinea and Melanesia,but also in the South Polynesianarea, supportsthe conclusionthat its introductioninto these regionsis due to an early Indonesianinfluence. On the westernborder of Indonesia the stick attachmentis associatedwith a singleoutrigger of several boomrls in tbe Andamans,as it is in Melanesia generally. Althoughthe Andamaneseare certainlyamong the most primitiveof all existingpeoples, there are indicationsthat they have borrowedcertain elements of a higherculture, of whichthe outriggercanoe is one. In the Nicobarsthe same type of stickattachment is associatedwith a singleotutrigger of two booms. 2. The rodattachment appears to be confinedto Indonesia (p. 87). 3. The spikeattachment is somewhat similar to the foregoingand is characteristic of and I believe peculiarto GeelvinkBay in New Guinea (p. 79); it consistsof a thin stickor spike which is driven throughthe boom and into the float; usually a thin branch extends at rightangles fromthe spike whichis lashed on to the boom. Friederici(II, p. 251) says that thisattachment is moreeasily taken to piecesthan the Moluccanor Halmaheran,and thatthe nativeshang up the hulls of the canoes in the corridorof the turtle-roofedhouses. A spike attachment,combined with a bowed spar whichpasses along the top of the end of the boom to whichit is lashed,the other end being lashed to the float, occurs at Nukutavaka, Paumotu Group (Alexander,pp. 766, 767). 4. The y-shaped stick attachment,though not recordedfor Indonesia, occurs widelyin the West Pacificand is characteristicof SouthernMicronesia, but it appears to be absent fromthe Mariana(Ladrone) and Marshallgroups. In the Pelew Islands each boom maybe supportedin the forkof one or two of thesesticks, it is prevalent in the Carolinesand occursin the Gilbertsand at Nauru (Fig. 34). It is lashed to the side of a boom in Liueniua (OntongJava) and Sikaiana on the easternflank of the Solomons. In the Liueniua canoe recordedby Friedericithe stem of the Y is immersedin the floatand thereis a lashingin addition,so that superficiallyit some- what resemblesthe U -Moluccanattachment. A double overcrossedY-stick occurs in the Loyalty Islands and New Caledonia. Its distribution,therefore, is W.-E., roughlybetween 100 N. lat. and the Equator, and roughlyN.W.-S.E. fromLiueniua

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 128 A. C. HADDON.-The Outriggersof Indonesian Canoes. and Sikaiana to New Caledonia and the Loyalty Islands, but is unrecordedelsewhere (cf. Haddon, 1918,No. 68).

FIG. 34.-ATTACHMENT OF A CANOE, ekuo, NAURU, MICRONESIA, HORNIMAN MUSEUM (cf. HAMBRUCH, 1915,fig. 252).

5. The boardattachment is typical of Indonesia; in thisthe boom passesthrough a board whichis insertedinto the float (p. 87). A preciselysimilar arrangementis foundon the coast of East Africa,but witha stickor blockof wood insteadof a board (Haddon, 1918,No. 29). Here it is associatedwith a double outriggerof two booms. The East Africanoutrigger canoe is universallyrecognised as having been derived fromIndonesia, perhaps, as Hornellsuggests, before indirect attachments were lashed to the floatin the latter region. The insertedstick attachmentis thus confinedto marginalareas. A somewhatsimilar arrangement occurs at Wukuhiva,Marquesas Islands, Alexander(p. 745) figuresa canoe with a singleoutrigger, each of the two booms of whichpasses througha board, the lower end of which is fastenedto the float, Accordingto his description,the ends of the booms " are seized to the perpendicularpieces, or stanchions,the lengthof whichis, as.a rule, the distance fromthe gunwaleto the waterline. These pieces are seized to the top side of the float[and not insertedinto it as in a true board attachment]. . . The crosspieces [booms],float, and stanchionsare braced withwithes." A curvedbrace is present. 6. The Balinese attachmentis confinedto the Eastern Javan area, but an in- terestinganalogy to it occursin Funafuti; Hedley says (p. 286) that the booms " are usually entire,but are sometimesmade divisible,spliced in a lock-joint[scarfed] and served. The advantage of detachingthe outriggerfloat fromthe hull occurs whenthe canoes are beachedand rolled over,the separatedhull beingmore-manage- able." The Balinese is really a directinserted attachiment, the Funafutiis a direct lashed one, but the occasional unshippingof the ends of the threebooms whichare permanentlyattached to the floatis a new,and so faras I know,unique, feature. In theSanta CruzIslands thewhole outrigger apparatus may be detachablefrom the hull.

(b) AttachmentTied to the Float. Lashed indirectattachments seem to have oustedthe inserted stick attachments throughouttthe greater part of Indonesia,probably because lashingthe attachment to, instead of simply insertingit into, the float was a more secure method of fastening,and at the sam-etime supplieda certainamount of elasticity.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 129

1. The Halmaheranattachment is confinedto Indonesia and the north-westof New Guinea. It varies considerablyin formand is widelyspread withinthe area, perhapsbecause it is adaptable to vesselsof large as well as of small size. 2. The Moluccanattachment has a restricteddistribution within the area (in my paper in Man, 1918, I erroneouslystated on pp. 117, 119 that it occurredin the Sulu archipelago). 3. The doubleor crossedU -Moluccanattachment occurs among the Barriai,Kobe,. and Kilenge on the northcoast of the westernend of New Britain,at Witu (French Islands) northof New Britain,and among the Nakanai of the northcoast of New Britainclose to the Gazelle Peninsula,at San Cristovalin the Solomons,and in the Tongan Islands (Haddon, 1918,No. 68). I was misinformedthat this attachmentis now obsoletein the Tongan Group,as A. G. Mayergives a photographof it at Vavau (1916, p. 25). With the exception of the last, which looks as if it indicated a definiteand probablylate culturaldrift, it is significantthat no indirectattachment tied to the floathas been reportedfrom Oceania, the inferencebeing that thismethod developed in Indonesiaafter all the greatnmigrations had takenplace. (The introductionof the moninto the Solomon Islands is, however,of relativelyrecent date.) If the Indo- Javaneseships traded between Java and India we may assume that theyalso traded in Indonesia,and thusthe knowledgeof an effectivetied attachmentshould have been knownthroughout the area. If theseships had a Halmaheranattachment we could assertthat the greatvoyages from Indonesia to Oceania musthave taken place long beforeA.D. 600, for if this attachmentwas suitable for ocean-goingships it would surelyhave been taken into the Pacificand have persistedsomewhere there. The entire absence of double canoes from Indonesia and their occurrencein Oceania suggeststhat thesecraft belong to an old culturestratum and one whichwas contem- poraneouswith the insertedstick attachment, at all eventsthese are at presentasso- ciated togetherin Oceania and New Guinea. The Hawaiian type, with its single outriggerand directlashed attachment,may belongto a special migration. The persistenceof the double outriggerin Indonesia requiressome explanation,. and it seemsto be due to the fact that whenprovided with an outriggerthe dug-out canoeis a veryhandy, light and stable craft,which is sufficientfor the general purposes of an essentiallyfishing community. For more extensivetrading voyages boats built up of planks were constructed, probablyunder foreign influence, of the Moluccan orembaitype. There can be no doubt that these at firstretained the double outrigger,as the korra-korrastill do;. Friedericisays thisis a craftof the constructionof a largeorembai, but withoutriggers (II, p. 237). In Indonesiait was foundthat the outriggerwas unnecessaryfor large- sailing craftand so it was discardedin the orembai. Perhaps the large waga of the Massimdistrict of BritishNew Guinea werederived from vessels of this type before the outriggerwas discarded,but in the case of the uaga the outriggeris single. VOL. L. K

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 130 A. C. HADDON.-The Outriggersof Indonesian Canoes.

Friedericipoints out that the mrn (mdnna,mnrd, m&l4, etc.) type of built-up canoe ofthe SolomonIslands, whichis withoutan outrigger,is constructedin a similar mannerto the orembai,indeed, Tasman comparedthe plank-builtmon of South New Ireland to the " corre-corre" of Ternate (Fr., II, p. 238). The monof the Solomon Islands have been developedon speciallines, and it is open to discussionwhether they were derived from an outriggerlessplank boat or whetherthe outriggerswere discardedin these islands. Friederici(III, p. 160) says that the monis not a good sea-goingcraft (and he says the same is true of the orembai);it is truethe Solomon Islandersmake long voyages in them,but theyare verycareful to choosegood weather. He is inclined to attributeits introductionto the second branch of his Alfuran nmigrationand not-to the Philippinemigration (cf. Man, XVIII, 1918, p. 118). A criterionof the Philippinemigration is the termguban (a Sulu archipelagoname {or a canoe with a double outrigger),but in the West Pacificthe double outrigger is retainedonly in the kop (a variantof guban)of Nissan. We have now to attemptto determinewhere the outriggerwas invented. As outriggercanoes are at the presenttime absent fromthe mainlandof Asia,' except in the SouthernIndian area, it is highlyimprobable that the SouthernMongo- loids (Proto-Malays)brought this craftwith them, though it is equally probablethat earlierimmigrants invented the outriggerafter they had establishedthenmselves in the islands. I have alreadypointed out (p. 126) that it is veryimprobable that this inventionwas due to Negritosor Papuans, and I may here add to these the un- progressivePre-Dravidians. This leaves us with the Indonesians and that early admixtureof Indonesianswith the firstswarm of Proto-Malaysto whichthe term Alfur has been applied (p. 109), and we inay thereforeconfidently attribute the invention to them. Probably later immigrantsadopted the outrigger,but the Orang Malayu always seem to have preferredthe outrigger-lesscraft to whichthey were accustomed. Indeed it may very well be that this device was due to the inventivenessof the Indonesian, rather than to the Mongoloid,element in the Alfurs. The early Arab navigatorsand otherswho may have come fromthe Red Sea or PersianGulf were ignorant of thiscontrivance, as werealso the Chinesenavigators. Thus of all the voyagersto the East Indian archipelago,the onlypossible introducers of an outriggerwere Indians, and it is doubtfulwhether they originally employed it. In certainlarge areas of Indonesia outriggercanoes are extremelyrare or even absent altogether. The main reason for this absence seemnsto be a knowledgeof the art of buildingsea-going plank boats, an art whichwithout doubt was introduced I Mr. I. H. Evans, of Perak,informs-me that his Malayassistant told him that a simpletype of doubleoutrigger is occasionallyfitted temporarily to Malay boats,generally to theprahu sagor, a small dug-out,when heavilyladen. This is the only recordknown to me, and it does not invalidatethe generalizationhere made, as the idea may have been borrowedfrom the islands. Mr. Evans has " seen large treetrunks, whose wood was of greaterspecific gravity than water, lashed to the sides ofprahus, something like outriggers,in bringingthem by sea."

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof IndonesianCanoes. 131 fromwithout, and to whichArabs, Indians, and Chinesehave contributedin varying degreesat differenttimes; primitivelythis art appears to have come fromAncient Egypt. An analogous change has taken place in the Solomon Islands owingto a cultural driftfrom the Moluccas (Fr., II, p. 161). An inspectionof the map on p. 71 suggeststhat the presentfocus of outrigger canoes is in the Moluccas,and it is legitimateto supposethat fromIndonesia, if not actually fromthe Moluccas, migrationstook place at various times,each with its specialtype of canoe or withsome partial modification. As a generalrule one might expectto findthat the earliertypes of canoes or of outriggerswere those that went furthest,and those that startedlast would have a miorelimited distribution; but we mustalso rememberthat the later swarmswould be morecivilised and have a better technical equipment,and thus some of them may have passed over earlierlayers and have reacheda fardestination. The generaldistribution of the maLntypes is as follows:- Marginal: Double outriggerwith two booms and a stick or direct lashed attachment; but the double outriggeris also characteristicof the most centralarea. Single outriggerwith two booms and stick attachments: north coast of BritishNew Guinea and South Polynesianarea. Single outriggerwith several booms and stick attachments: Nico- bars, Andamans,parts of New Guinea,Melanesia. Single outriggerwith two booms and a directlashed attachment: South Indian area and North Polynesianarea. WVithinthe area: Widelydistributed, double outriggerwith two booms and a direct lashed attachment. Less distributed,double outriggerwith two booms and a Halma- heran attachment. Most restricted,double outriggerwith two booms and a Moluccan attachment.

If we apply the principlesenunciated at the beginningof this sectionwe are led to concludethat the above orderroughly represents an historicalsequence in which it seems evidentthat, of the indirectattachments, the stick is the oldest and the Moluccan the most recent. There also appears to be a probabilitythat the double toutriggeris moreancient than the single,despite its persistencein Indonesia,a fact -whichdoes not fitin withthe foregoingargument.

BIBLIOGRAPHY. Adriani,N., and Kruyt,A. C.-De Baree-sprekendeToradjas van Midden-Celebes.Platen en Kaarten. s' Gravenhage,1912. Alexander,A. B.-" Notes on the boats, etc. of the South Sea Islands." U.S. Com. of Fish and Fisheries. Part xxvii,Report of the Commissioner for 1901,Washington, 1902. K 2

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 132 A. C. HADDoN.-The Outriggersof Indonesian Canoes.

Anson,G.-A VoyageRound the World. 1748. Baessler.-Internat.Arch. f. Ethnogr.,IV, 1891. Beeckman,D.-A Voyageto and fromthe Island of Borneo. 1718. Bickmore,A. S.-Travels in theEast Indian Archipelago. 1868. Brenner,J. F. von.-Besuch bei den KannibalenSumatras. Wiirzburg,1894. Brown,J. Macmillan.-Man., XIV, 1914, No. 43. Bry, J. T. de.-India Orientalis,Pars V. (Quinta Pars, IndiceOrientalis.) Francofurti,1601. Burbidge,F. W. T.-The Gardensof the Sun. 1880. Christian,F. W.-Eastern PacificLands: Tahiti and theMarquesas Islands. 1910. Clereq,F. S. A. de, and Schmeltz,J. D. E.-EthnographischeBeschrijving van de West-noordkust van NederlandschNieuw-Guinea. Leiden, 1893. Dumont d'Urville,J. S. C.-Voyage pittoresqueautour du Monde. Paris, 1834. Edye, J.-Journ. Roy. Asiat. Soc., I. London, 1834. Ellis, W.-Polynesian Researches. 2nd Edit., 1831. Fischer,H. W.-Katalog Ethnogr.Reichsmuseums. X., Miittel-Sumatra(Sumatra, III). Leiden, 1916. Folkard,H. C.-The Sailing Boat. 5th Edit., 1901. Forbes, H. O.-A Naturalist'sWanderings in theEastern Archipelago. 1885. Forrest,T.-A Voyageto New Guineaand theMoluccas, etc., during the years 1774, 1775 and 1776. London,1779. Voyageaux Moluqueset a la NouvelleGuinge. Paris, 1780. Friederici,G. (referredto as Fr.).-WissenschaftlicheErgebnisse einer amtlichen Forschungsreise natchdem Bismcarck-Archipelim Jahre1908. II.-" Beitriigezur Volker-und Sprachenkundevon Deutsch-Neuguinea." Mitteil.aus den Deutsch.Schutzgeb., Ergiinzungsheft Nr. 5, 1912. III.-" Untersuchungenuiber eine melanesischeWanderstrasse." Ibid. Erg.-heft,Nr. 7, 1913. Fritsch,G.-Globus, XCI., 1907. Giglioli,H. H.-Int. Arch.fu,r Ethnogr., VI., 1893. Gomes,E. H.-Seventeen Years amongthe Sea Dyaks of Borneo. 1911. Graafland,N.-De Minahassa. Haarlem, 1898. Graebner,F.-Zeitschr. f. Ethnologie,XXXVII, 1905. -- - Anthropos,IV., 1909. Grubauer,A.-Unter Kopfjcgernin Central-Celebes.Leipzig, 1913. Guillemard,F. H. H.-The Cruiseof the Marchesa. 2nd Edit., 1889. " Australasia." II. Standford'sCompendium of Geography.1894. G.-Photographs in the possessionof Dr. Guillemardtaken when on the cruise of the " Marchesa,"1881-1884. Haddon, A. C.-The Wanderingsof Peoples. Cambridge,1911; 2nd Edit., 1920. " The outriggercanoes of TorresStraits and NorthQueensland." Essays and Studie& presentedto WilliamRidgeway. Cambridge,1913. - Man, XVIII, 1918,Nos. 29, 68. Hambruch,P.-Ergebnisse d. ,SidseeExped. 1908-1910,II, Ethnogr.B. MilkronesienBd. 1 Nauru, 2 Halbband. Hamburg,1915. Havell, E. B.-Indian Sculptureand Painting. 1908. Hedley,C.-The Atollof Funafuti. MemoirIII. AustralianMuseum, Sydney, 1897. Hickson,S. J.-A Naturalistin theCelebes. 1889. - Proc. Camb.Antiqu. Soc., VII (N.S., I.), 1891 (1893). Hornell,J.-Man, XIX, 1919, No. 55. Hose, C., and McDougall,

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions A. C. HADDON.-The Outriggersof Indonesian Canoes. 133

Kloss, C. B.-In the Andamans and Nicobars. 1903. Kramer, A.-Die Samoa-inseln. II. Stuttgart, 1903. Hawaii, Ostmikronesienund Samoa. 1906. Kor. Blatt d. Deutsch Ges. f. Anth. LXIII. 1911. Krause, F.-Jahrbuch d. stddl.Mus. f. Vdlkerklundezu Leipzig. I. 1906 (1907). Krieger, M.-Neu-Guinea. Berlin, 1899. Kubary, J. S.-Ethnographische Beitrdgezur Kentniss des Karolinen Archipels. Leiden, 1895. Kiukenthal, W.-" Forschungsreise in den Molukken und in Borneo." Abhandl. d. Sencken bergischennaturfors. Gesellschaft, XXII. Frankfurt-a-M., 1896. L., G. M. A. W.-Premier Livre de l'Histoire de la Navigation aux Indes.orientales,etc. C. Nicolas Amstelredam,1598. Landor, A. H. Savage.-The Gems of theEast. 1904. Lane-Fox (Pitt-Rivers), A. H.-Journ. Anthrop. Inst., IV, 1875, p. 399; reprinted in Evolution of Cultureby Pitt-Rivers,A. L. F., Oxford,1905. Journ.Anthrop. Inst., VII, 1878,p. 434. Langen, H. G.-Key- oder Kii-Inseln des 0.1. Archipel. Wien, 1902. Leemans, C.-B6ro-Boedoer, Leiden, 1873; B6r6-Boudour (French translation). Leide, 1874. Linschoten (Lintscotus), J. H. van.-in Bry, J. T. de, India Orientalis, Pars III (Tertia Pars, Indice Orientalis). Francofurti, 1601. Mager, H.-Le Monde polynesien. Paris, 1902. Man, E. H. Journ.Anthrop. Inst., XI, 1882,p. 268; XII, 1883,p. 117 ; XV, 1886,p. 428. Martin, K.-Reisen in den Molulkken. Leiden, 1894; Geologische Theil, 1903. Mayer, A. G.-" The Islands of the Mid-Pacific." The ScientificMonthly. II. New York, 1916. Meyer, A. B., and Richter, O.-" Geisterfallen im Ostindischen Archipele." Abhandl. u. Ber. des Konigl. Zool. u. Anth.-Eth.Museums zu Dresden. Bd. X. No. 6. Ethnograph.Miszellen. II. Berlin, 1903. Modigliani, E.-Un Viaggio a Nias. Milan, 1890. - L'Isola della Donne. Milan, 1894. Mouat, F. J.-Adventures and Researchesamong theAndaman Islanders. 1863. Muller, W.-Baessler-Arch., II, 1912, p. 235. Yap, Ergebnisse der SiMdsee-Expedition,1908-1910. Hamburg, 1917. Munday, R.- Yarrativeof Events in Borneo and Celebesfrom the Journal of James Brooke. 1848. Nicolas, C. [Printer].-Le Second Livre Journal ou Comptoir,contenant le vrayDiscour8 et Narration historiquedu Voiage faict par les hiuct Navires, etc. Amsterdam, 1601. Nieuenhuis, A. W.-Quer durch Borneo. Leiden, 1904-07. Pallander, E.-The Log of an Island Wanderer. 1901. Partington, J. Edge, and Heape, C.-Ethnographical Album of thePacific Islands. 1. Manchester, 1890. Perouse, Atlas du Voyage de la-London, 1798. Perry, W. J.-The Megalithic Culture of Indonesia. Manchester and London, 1918. Pfluger, A.-Smaragd-inseln der Si6dsee. Bonn., n.d. Pritchett, R. T.-Pen and Pencil Sketchesof Shipping and Craftround the World. 1899. Radhakumud Nookerji.-Indian Shipping. London, 1912. Riedel, G. F.-De sluik-en kroesharigerassen, etc. s' Gravenhage, 1886. Rivers, W. H. R.-The History of Melanesian Society. Cambridge, 1914. Rosenberg, H. von.-Der MalcayischeArchipel. Leipzig, 1878. Int. Arch. fiurEthnogr., I., 1888, p. 218. Roth, W. E.-" North Queensland Ethnography." Bull. 14. Records of the Australian Museum, VIII. Sydney, 1910. Roth, H. Ling.-Crozet's Voyageto Tasmania, New Zealand, etc. 1891. Safford,W. E.-" Guam and its People." Smithsonian Report, 1902 (1903). Sarasin, P. and F.-Reisen in Celebes. Wiesbaden, 1905. Smith, V. A.-The early Historyof India, 3rd Edit. Oxford, 1914.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions 134 A. 0. HADDON.-The Outriggersof Indonesian Canoes.

Svoboda,W.-Int. Arch.fitr Ethnogr., V, 1892; VI, 1893,1. Thilenius,G.-" Ethnograph. Ergebnisse,aus Melanesien1. Theil,Die polynesischenInseln an der OstgrenzeMelanesiens." Nova Acta. Abh.der Kaiserl. Leop.-Carol.Deutsch. Akad. der Natturforscher,Bd. LXXX, Nr. 1. Halle, 1902. Valentijn,F.-Oud en Nieuw Oost-Indien,IV. 1726. Vojnich,O.-The Island Worldof the Pacific. Budapest,1909. Wallace, A. R.-The Malay Archipelago. 1869. Weule,K.-Leitfaden der Volkerkunde.Leipzig, 1912. Wilkes,C.-Narrative of the United States Exploring Expedition. Londonand Philadelphia,1845.

ADDENDA. Friederici,G.-Malaio-Polynesische Wanderungen. Leipzig, 1914; 'Verhandi.des XIX DeutschenGeographentages zu Strassburgi. Els. 1914. Berlin,1915. These reachedme aftermy paper was in print; the earlierof thesetwo papers is the more valuable as there are copious references.The author believes that the Malayo- Polynesianmigrations left Indonesia before the Hindu influencethere, and tookplace in the secondand fourthcenturies A.D., and thatthe Sumatrian migration to Madagascartook place in thetenth century A.D. He suggeststhat the Malayo-Polynesian migrations were originally made on raftsof threebeams, and that canoes withdouble outriggerswere developed from them; the reductionof theseto singleoutriggers " is quite a naturalprocess " ; the double canoe arose fromthe latterbv the floatbeing increased in size and hollowedout. Hornell, J.-" The Origins and EthnologicalSignificance of Indian Boat Designs." Mem. AsiaticSoc. ofBengal, VII, p. 139, Calcutta,1920. This is a veryvaluable monograph. Hornelldiscusses the Javaneseships of the Boro- budur sculpturesand gives new drawingsof my Figs. 25 and 28 which clear up certain points; both have biped masts,the latterhaving rungs between the limbs,and the three booms of this ship are distinctlycurved and pass betweenthe two elementsof the float. " Les Pirogues a balancierde Madagascar et de l'Afriqueorientale." La G&ographie, XXXIV. Paris, 1920,p. 1. "Madagascar and East Africa: Canoes." Man, 1920,No. 67. These excellentpapers have appearedtoo late forme to referto them,the formeris the longerand has moreillustrations. Hornellshows that the followingvarieties occur on the west coast of Madagascar,all with two booms: 1. Direct insertedin a modelof a sailingboat withan ordinaryfloat on one side, and on theother the booms carry a reducedfloat which evidently serves as an outlayer. 2. Double outriggerwith a smallboard (" peg stanchion") attachmentsto thetwo floats. 3. Singleoutrigger with similar attachments, the boomson the otherside beingtied to a lightbamboo spar to forman outlayer. 4. -do.-do.-, but onlythe fore boom is prolongedon theother side, it servesto extend the starboardsheet. In the sixteenthand seventeenthcenturies all the canoes appear to have had douible outriggers,by the middleof the nineteenthcentury both formsoccurred, now none but singleones occur. The attachmentin Madagascaris alwaysvertical, as it is at Lamu, butin theother East Africancanoes it is oblique and the flatboard is similarlyhorizontal or canted. Hornell rightlyobserves that thesefacts justify his previousconclusion that the outriggercanoes of Madagascarand East'Africaare deriveddirectly from Javanese [or at all eventsIndonesian] types. He pointsout that " In theirconsistently,double form, the Africanvarieties have retaineda primitivestructure almost entirelylost in present-dayMadagascar outriggers " and that " the Lamu varietyhaving vertical stanchionsapproaches most closelyto the Madagascarform." I may add that I have just comeacross a verybadly drawnillustration of a canoe from Mohelia,with a doubleoutrigger; each pair ofthe two divergent booms appear to be directly connectedwith its float. T. H. [Herbert]Esquier, A Relationof Some Yeares Travaile, BegunneAnno 1626; London 1634.

This content downloaded from 195.78.108.81 on Tue, 24 Jun 2014 21:31:00 PM All use subject to JSTOR Terms and Conditions