Volume 98 1989 > Volume 98, No. 3 > Wait for the west wind, by B. Finney, p 261-302
                                                                                             Previous | Next   

- 261

Why were the islands along the eastern edge of Polynesia settled by seafarers who, many generations removed, ultimately stemmed from the faraway Asian side of the Pacific, and not by people from the much closer shores of North and South America — particularly when interisland distances increase, and sailing conditions become more difficult, the further into the ocean one moves from the Asian side? Keegan and Diamond (1987) propose that island distribution, not sheer proximity, determined the direction of settlement: Pacific Island colonisation proceeded from west to east because, starting immediately offshore the Asian mainland, a virtually continuous series of island chains encouraged migrants to sail further and further out into the ocean by rewarding them with island after island on which they could settle, while, in contrast, the open ocean between New World shores and Polynesia provided no such opportunity and incentive for those who lived along the Pacific coast of the Americas. To account for the ability of succeeding generations of migrants to cross the increasingly wider interisland gaps — which jump from the tens of miles in the south-western Pacific, to the hundreds of miles in the central Pacific and then reach over a thousand miles in the far reaches of Polynesia, Green (1978:3; MS 1987b) hypothesises that, during this long migration, the double-canoe was developed into a stable and seaworthy vessel to carry all the people, their provisions, and their breeding stock of plants and animals needed to establish colonies on the increasingly distant and resource-poor islands. This paper addresses the question of how the Polynesians were able to continue this migration so far into the southeastern Pacific when they had to sail against the direction of the trade winds which generally blow from between east and south-east and which become increasingly dominant the further east one moves across the ocean.

To cite an extreme view of the problem of moving eastwards against the trade wind direction, Heyerdahl (1978:332) maintains that Polynesia could not have been colonised from the west because “the permanent trade winds and forceful companion currents of the enormous Southern Hemisphere” would have prevented canoe sailors from the Asian side of the ocean from - 262 ever sailing to the east. If the trade winds throughout the South Pacific were permanent, it would indeed have been very difficult to sail a canoe directly eastwards over long distances. To be sure, Polynesian double-canoes are excellent sailing vessels. When sailing on a beam reach (with the wind blowing at right angles to the hull) or a broad reach (with the wind blowing abaft the beam, or greater than 90° to the hull), they move easily and swiftly through the water (Figure 1). However, while double-canoes can sail to windward, they cannot do so as well as a yacht equipped with a deep keel or centreboard. In fact, to attempt to sail a shallow-draft, keel-less vessel like a double-canoe too close into the wind can be self-defeating. As it points closer and closer into the wind, a double-canoe slows noticeably and begins to make so much leeway that little progress can be made directly into the wind, particularly when sailing against a strong current.

Instrumented sailing trials with the reconstructed Hawaiian double-canoe Nālehia, and the 1976 voyage of Hōkūle'a from Hawai'i to Tahiti made by sailing diagonally across the trades, indicate that a double-canoe sails most efficiently to windward when it is sailed “full and by”, which means holding the canoe as close to the wind as possible without making too much leeway or losing the full drive of the sails. Sailing full and by against a 15 to 20 knot wind, a double-canoe can do at least 4 to 5 knots and “make good” a course of about 75° off the wind (calculated by measuring heading against the true direction of the wind, then adding leeway) (Finney 1977). Because a double-canoe cannot point significantly closer to the wind without greatly losing efficiency, it must make long, shallow tacks to one side and then the other of a windward goal, which means it must be sailed almost four miles for every mile made directly to windward (Figure 2; Finney 1985:10). A 500-mile voyage made tacking directly to windward would, therefore, be almost 2,000 miles long in actual sailing distance, and a 1,000-mile voyage almost 4,000 miles long. Struggling against the wind-driven current would, in effect, increase these sailing distances even further. Although the seaworthiness and agility of the ancient canoes were undoubtedly crucial to the voyaging accomplishments of the Polynesians, it is unlikely that they or their ancestors ever made long voyages solely by tacking directly to windward — particularly in vessels slowed by a large complement of men, women and children, as well as a heavy load of tools, supplies, plants and animals.

However, the exact degree to which Polynesian craft could sail to windward is not the main issue. No sailor, even one with a modern yacht of superlative windward characteristics, wants to tack long distances to windward when he can go where he desires by sailing before the wind. Transoceanic sailors, for example, typically seek out zones of favourable winds for

- 263
Best points of sail for Hōkūle'a (cf. Kane 1976:110).
Distance a yacht and a double-canoe must travel tacking against the wind.
- 264

crossing the seas. When the Spanish sailed their famous Manila Galleons from the Philippines to Mexico, they did not try to tack east against the northeast trades. Instead, they first headed north until after they had left the tropics and the trade winds and encountered steady westerlies of the higher latitudes. They then turned east to sail before the westerlies until sighting land somewhere along the North American coast. It is possible that, once the main islands of tropical Polynesia had been settled, some adventurous canoe voyagers wandered far enough to the south of the Austral Islands or north of Hawai'i to encounter steady westerlies, and then used these to explore unknown seas further to the east (Finney n.d.). However, those pioneering voyagers who discovered and settled the islands strung across the South Pacific from South-east Asia to Polynesia had a shorter, warm-weather, alternative to lengthy, roundabout voyages made via the higher, and colder, latitudes: they could stay in the tropics and wait for the west wind.

Not only did Heyerdahl have the direction of settlement wrong, but he also attributed a permanence to the trade winds which does not exist. The error of

Stylised Austral winter wind circulation in the South-west Pacific (after Hessell 1981:43; Steiner 1980:9).
- 265

assuming that trade winds are permanent, or nearly so, is not uncommon. Trade winds are typically defined as steady, regular winds that blow from an easterly direction in tropical latitudes. They have come to be known as such “from the great regularity with which they blow, thus assisting the ships which used to carry trade around the world in the days before steam propulsion” (Kemp 1976:882). Globes and maps upon which trade winds are indicated by curving arrows marching from east to west across tropical seas give graphic expression to the apparent steadiness and regularity of these winds. In reality, however, although the trades are the most steady and regular of global winds, they none the less wax and wane, and in some regions may be absent for days, weeks or even months at a time.

In that part of the South Pacific traversed by those seafarers who were ancestral to the Polynesians, the winds have a marked monsoonal pattern. Whereas the trades may blow fairly steadily over the seas between South America and the Tuamotu Islands traversed by Heyerdahl aboard the raft Kon-Tiki, there are two distinct wind seasons along the actual migration trail

Stylised Austral summer wind circulation in the South-west Pacific (after Hessell 1981:42; Steiner 1980:9).
- 266

of the ancestral Polynesians from South-east Asia to the western edge of Polynesia: the Austral (Southern Hemisphere) winter when easterly trade winds generally prevail, and the Austral summer when westerly winds regularly intrude (Hessell 1981; Revell 1981; Steiner 1980; Thompson 1986a, 1986b). Firth (1936:30), for example, considers that the principal climatic feature of Tikopia, a Polynesian Outlier located in Melanesia off the eastern end of the Solomon Islands, is the marked seasonal difference between the period Tikopians call tonga from April till September when easterly trade winds prevail, and that of raki from October till March when the trades are largely replaced by variable westerlies.

Figure 3 shows an idealised picture of wind circulation in the southwestern Pacific during the winter when easterly trade winds generally prevail clear across the tropical South Pacific. Figure 4 portrays how this easterly flow is typically interrupted in the summer by westerly winds that prevail in the south-western tropical Pacific, and extend eastwards into mid-Pacific waters. This monsoonal pattern arises from the summer heating of the Australian continent and the formation of a trough of low pressure extending over northern Australia and eastwards out into the Coral Sea. Although, on the southern side of this monsoon trough, the winds continue to blow mostly from the south-east, they frequently blow from the west and the north-west on its northern flank, flowing towards the trough.

The influence of this monsoon trough extends far into the Pacific, although the westerlies generally become more and more episodic the further east one sails. Whereas they may dominate during the summer months along the north shore of New Guinea and adjacent islands, they may only infrequently penetrate as far east as the Marquesas Islands. The arrows in Figure 5, which stand for mean surface wind direction and velocity for the month of January as calculated from ship reports collected since the last century, clearly show a corridor of westerly winds extending east from New Guinea almost to Fiji. Beyond Fiji the westerlies become progressively more episodic, and affect the mean wind flow depicted in the figure only by giving the arrows a slight northerly slant. Yet, these bursts of westerlies can dominate the seas in East Polynesia for days on end, as is indicated in Figure 6, which portrays the surface wind circulation during one such episode that lasted for slightly over a week during January 1979 and brought strong westerly winds as far east as Tahiti.

In some years, particularly when there is a strong E1 Niño, westerlies may extend further than usual into Polynesian waters and last longer. The flood of warm, equatorial water down the South American coast which characterises an E1 Niño is preceded by a marked change in the atmospheric pressure

- 267
Mean surface wind direction and velocity for the tropical Pacific during January (after Wyrtki and Myers 1975: Fig. 1a).
Surface wind analysis for January 16, 1979 (after Finney 1985:15).

gradient over the South Pacific and a consequent disruption of the trade wind field. (When the trade winds weaken, the sea surface, normally higher in the west than the east because of the stress of the trades, begins to level off, triggering the warm water flows that inundate the cold water upwelling off the Peruvian coast that nourishes the rich fisheries there.) The westerlies then prevailing in the western Pacific generally penetrate further east than the regular monsoonal westerlies, and may also be more persistent. For example, during 1982–3, when the strongest El Niño event recorded so far occurred, the westerlies began to reach the Marquesas in December, and were dominant there during the following March and April (Finney 1985; Sadler and Kilonsky 1983; Figure 7).

These westerlies — both the regular monsoonal type and those intensified

- 268
Surface wind analysis for March 21, 1983 (after Finney 1985:17).

and extended during El Niño events — must have facilitated the movement of people eastwards into the Pacific. Although we have no direct record of pioneering voyagers exploiting these winds, there is abundant evidence that later island sailors were well acquainted with the alternating rhythm of easterlies and westerlies, and waited for the wind appropriate to the direction towards which they wanted to sail. For example, in Indonesia, one of the few places in the world where significant amounts of freight are still carried by sailing craft, traders have for centuries exploited this monsoonal wind pattern to sail east along the Indonesian chain before the westerlies, and then, with the return of the trades, back to the west. Further into the Pacific, canoe sailors in surviving trading networks along the coast of New Guinea and among other Melanesian islands are similarly dependent upon seasonal alternations of wind patterns to facilitate their voyaging.

That traditional Polynesian sailors were familiar with the more episodic spells of westerlies common in their waters, and knew how to use them to sail to the east, is well attested in the writings of early European navigators, scientists and missionaries. (Banks 1962:1:368; Beechey 1831:1:251–2; Cook 1955:137,154; Hale 1846:118–9; La Pérouse 1798:3:66; Williams 1837:504–10). For example, when Captain James Cook first visited Tahiti in 1769 he was evidently puzzled by the tales the Tahitian Tupa'ia told about how the Tahitians voyaged back and forth to islands many days sail to the west. Cook realised that they could easily sail west before the trade winds, but - 269 he wondered how they could sail east, back to Tahiti, against those same winds. The learned Tahitian soon set Cook at ease, however, by explaining that they waited for summer spells of westerly winds to make their return: “Tupia [Tupa'ia] tells us that during the Months of Novr Decembr & January Westerly winds with rain prevail & as the inhabitants of the Islands know very well how to make proper use of the winds there will no difficulty arise in Trading or sailing from Island to Island even tho' they lay in an East & West direction” (Cook 1955:154).

The term “westerly winds” refers to any wind with a westerly component; that is, to winds from the north-west and south-west as well as straight out of the west. North-west or south-west winds are ideal for sailing due east, for a canoe generally sails best on a broad reach, with the wind coming over the stern at an angle, rather than directly from astern (Figure 1). Thus, a canoe heading due east can sail best broad-reaching with a north-west or south-west wind, not running directly before a wind straight out of the west. In fact, a canoe can still make good progress to the east reaching with a north wind, or south wind, or even with a wind slightly forward of either north or south.

On the basis of the periodic extension of westerly winds into Polynesian waters, and their documented use for sailing eastwards by Polynesian mariners of the early contact period, Finney (1977:1283; 1985) has proposed that the pioneering voyagers who first sailed from west to east across Polynesia did so by exploiting westerly wind episodes to sail east, tacking to windward only when and if the westerlies dropped and were replaced by easterly trade winds. In mid-1986 we had a chance to test the feasibility of that mode of passage by sailing the reconstructed Polynesian voyaging canoe Hōkūle'a from Samoa to Tahiti. Here we report that voyage and explore its implications for understanding the colonisation of Polynesia.


The voyage was planned to investigate experimentally how it would be possible to sail a double-canoe across Polynesia from west to east in order to shed light on the initial stages of the colonisation of East Polynesia. The settlement of Polynesia began over 3,000 years ago when Lapita seafarers, whom archaeologists trace back to the Bismarck Archipelago off the north coast of New Guinea, reached the then-uninhabited archipelagos of Fiji (including the Lau group), Tonga and Samoa (Green 1979). As they explored and settled the many islands of these mid-Pacific chains, local adaptation and cultural differentiation followed. On Tonga and Samoa, as well as a few smaller outlying islands, the distinctive cultural pattern ancestral to all Polynesian cultures developed. These islands, now known as West Polynesia, - 270 and not any distant continental shore, are, therefore, thought to form the true Polynesian homeland (Emory 1959; Green 1967, 1981). Through our voyage we wished to investigate the crucial first step in the dispersal of this oceanic culture throughout the vast Polynesian triangle: the movement from West Polynesia to the central archipelagos of tropical East Polynesia — the Cook Islands, the Society Islands (including Tahiti), the Marquesas Islands, the Austral Islands, and the Tuamotu Islands.

In April of 1986 Hōkūle'a was sailed from the Bay of Islands, in the North Island of New Zealand, first to Tonga and then on to Samoa in order to position the canoe for an attempt to sail from Samoa to Tahiti, via Rarotonga. By sailing from Samoa to the East Polynesian islands of Rarotonga and Tahiti we did not intend to imply that the initial migration, or migrations, to East Polynesia necessarily went that way. Although Samoa is a leading candidate as a dispersal point, voyages to the east may also have been made from Tonga and possibly from one or more of the smaller islands of West Polynesia as well. Besides, because we obviously already knew the location of Rarotonga and Tahiti, as well as that of all the other East Polynesian islands, we were hardly in a position to recreate an early discovery voyage. That would have required that we strike out randomly to the east, without any foreknowledge of what archipelagos and islands might be there. In fact, our choice of East Polynesian landfalls was, in large part, dictated by other reasons. We had sailed to New Zealand via Tahiti and Rarotonga, and were obliged to call upon these islands on the way back in order to fulfil pledges made to Tahitians and Rarotongans on the way to New Zealand.

None the less, despite our geographical knowledge and port-of-call obligations, we believed that, in attempting to cross the ocean gap between Samoa and Tahiti, we could learn much about the sailing problems faced by early voyagers who headed east from the West Polynesian homeland, and how these may have affected the settlement of East Polynesia.

Upon Nainoa Thompson, the navigator of Hōkūle'a, fell the responsibility of determining the sailing strategy to be followed. The task was formidable because a course straight from Samoa to Rarotonga and then on to Tahiti would take the canoe almost directly against the direction of the prevailing trade winds. After reviewing the wind patterns, and considering what had been learned from sailing Hōkūle'a over the previous 10 years, Nainoa concluded that it would be extremely difficult to try to tack to Rarotonga and Tahiti against the trades, and that the best hope of reaching these islands would be to exploit favourable wind shifts to make easting. However, although Nainoa agreed that the Austral summer, with its episodes of monsoonal westerlies, would be the time of year when we would most likely be able to - 271 find favourable winds for sailing east, he declined to schedule a summer sailing. These bursts of monsoonal westerlies, Nainoa noted, are often blustery and accompanied by heavy cloud cover and rain. Furthermore, the summer period, when westerlies are most prevalent, is also the hurricane season, when tropical disturbances are most likely to cross the route between Samoa and Tahiti, particularly during major El Niño events such as that of 1982–3 when four tropical storms did so. Accordingly, for the safety of crew and canoe, Nainoa ruled out applying the strategy of using summer westerlies to sail east.

Instead, he proposed a variant of that strategy. Upon studying the daily meteorological charts for the preceding several years, Nainoa and University of Hawai'i meteorologist Thomas Schroeder noted that, even during the Austral winter, when the trade winds are normally most dominant, brief spells of westerly winds occasionally interrupt the trades when low-pressure troughs emanating from subtropical depressions to the south reach into the tropics. This suggested the strategy of exploiting the westerlies of these passing troughs to sail east, and, when forced to sail against the trades, of dipping down into more southerly latitudes to be better positioned to catch the favourable winds associated with the passage of the next trough. 1 Accordingly, Nainoa proposed to make Hōkūle'a ready for a mid-winter departure, to wait for the coming of a low-pressure trough, and then to try to sail the canoe as far east as possible with the westerly winds that would accompany the passage of the trough. Once the trough had passed, Nainoa then proposed to tack against the trades while waiting for another trough in order to gain another boost to the east. Through a combination of exploiting the westerlies of passing troughs, and of tacking against the trades while waiting for the next trough, Nainoa hoped that Hōkūle'a would be able to work its way first to Rarotonga, and from there on to Tahiti. If, however, the troughs did not extend into the tropics that year, Nainoa knew that it would be a long and hard struggle to try to make easting by tacking against the trades, and that it might not be possible to reach Tahiti, or even Rarotonga.


The Austral winter of 1986 turned out to be highly favourable for sailing to the east, although we did not fully realise that until Hōkūle'a was well out to sea, and did not completely understand what had occurred until after the voyage when we could analyse the meteorological data for the South Pacific gathered by satellites and surface stations.

Nainoa's study of the meteorological charts of the South Pacific for previous years had revealed an uneven pattern. During the winter of 1984, for

- 272
Stylised wind circulation between Samoa and Tahiti during the winter of 1986. See Figure 9 for a key to meteorological symbols used. Arrows indicate general surface wind direction.

example, the trades broke down seven times with the passage of low-pressure troughs. During the winter of 1985, however, he could find no evidence of any troughs significantly interrupting the trade winds, which appeared to stay solidly in place all winter long. The first indications that the 1986 winter might be more like 1984 than 1985 came from daily meteorological charts of the south-western Pacific, consulted after the canoe arrived in Samoa, which indicated that a wholesale disruption of surface wind circulation patterns over a large part of the South Pacific was developing: low-pressure troughs were beginning to extend into the tropics, sweeping across the route the canoe would follow to Tahiti, bringing with them wind shifts that would enable Hōkūle'a to sail eastwards.

In the typical winter pattern, idealised in Figure 3, trade winds dominate the route between Samoa and Tahiti. These blow primarily from the south-east between Samoa and Rarotonga, and more out of the east from Rarotonga to Tahiti. These south-east trades and “divergent easterlies” usually meet in a cloudy region called the South Pacific Convergence Zone, which is typically located along a line that runs south-east to north-west from the Austral Islands between the Southern and Northern Cook Islands towards Kiribati. Well south of the trades, along the latitude of New Zealand's North Island, high-pressure systems move from west to east. To the south of these highs, low-pressure systems also move from west to east. Only infrequently do low- - 273 pressure troughs extend far enough north from these depressions to reach into tropical latitudes and interrupt the trades and bring westerly winds.

During the winter of 1986, however, the travelling highs were centred about 10° further south than normal and were persistent and slow moving. Pressures at their centres were also higher than normal. These displaced highs, and anomalous upper air conditions at the same latitude, blocked the normal passage of the low-pressure systems, shunting some of them to the north between 25° and 35° South latitude before they could continue their eastward movement. The troughs from these displaced lows, therefore, extended further north than usual, repeatedly reaching into trade-wind latitudes where vertical coupling between an unusually persistent and quasi-stationary upper air trough located between Tonga and Tahiti, and the displaced surface troughs, intensified surface disturbances. In addition, the South Pacific Convergence Zone was largely absent. The subtropical ridges of high pressure extending from the migrating highs were also displaced further north than usual, and tended to connect to the semipermanent high-pressure system centred to the south-east between the Austral Islands and South America. This combination of displaced troughs and ridges frequently interrupted the trades and gave rise to long periods of anomalous northerly and westerly winds (Figure 8).

To be sure, satellite photographs, daily weather maps and detailed weather records, as well as a growing body of meteorological theory, enabled us to plot the occurrence of these westerlies before departure, and to explain their origin. None the less, although early Polynesian mariners lacked such modern aids, they must have had a deep practical knowledge of weather matters that comes to people whose lives depend upon the sea. Certainly, the European explorers who took the time to become acquainted with Polynesian sailors praised their meteorological skills. For example, concerning the two Tahitians he took with him from Otahiti (Tahiti) to Oriayatea (Ra'iatea), the Spanish navigator Andia y Varela (Corney 1915:2:286–7) reported that:

What took me most in two indians [the Tahitians] whom I carried from Otahiti to Oriayatea was that every evening or night they told me, or prognosticated, the weather we should experience the following day, as to winds, calms, rainfall, sunshine, sea and other points, about which they never turned out to be wrong: a foreknowledge worthy to be envied, for, in spite of all that our navigators and cosmographers have observed and written anent [about] the subject, they have not mastered this accomplishment.

While their sea-level perspective and lack of knowledge of global weather

- 274
Surface weather map for July 7, 1986. This map, and subsequent weather maps, depict conditions at 1400 Cook Island Standard Time. Sea-level pressure is indicated in millibars in this and subsequent weather maps. Arrows indicate general surface wind direction.

dynamics might seem limited to us today, had such weather-wise Tahitians wanted to sail to the east during a winter such as 1986, they surely would have seized upon the unseasonable westerlies just as we did. Although, before the voyage, Nainoa enjoyed a “God's-eye” view of the weather furnished by satellite photographs and detailed meteorological charts, once at sea he was in the same situation as these Tahitians and other early Polynesian sailors who had only their observations of the wind, clouds and swells to guide their - 275 sailing decisions — and he probably acted more or less as they would have.


In early July Nainoa and part of the crew flew from Honolulu to American Samoa in order to join the canoe and the rest of the crew at Ofu Island in the Manu'a Group at the easternmost end of the Samoan chain. 2 Although they arrived just as a low-pressure trough was approaching, they were not able to ready the canoe for departure quickly enough to take advantage of the accompanying wind shift. By the time the canoe was ready, the westerly winds had passed and the trades were re-established.

None the less, Nainoa chose to leave then, setting sail on the afternoon of July 7, by which time a small high spreading eastwards along 25° South the day before had become embedded in the subtropical ridge extending west-wards from the seas south-east of Tahiti, bringing a flow of easterly winds north of the ridge axis and into the region where Hōkūle'a was sailing (Figure 9). With the east-south-east trades then blowing, the canoe could have been sailed either to the north-north-east or to the south. To sail north-north-east would have taken the canoe away from Rarotonga and eventually out of range of any troughs extending up from the south. Nainoa's strategy was to make a long tack to the south both to approach the latitude of Rarotonga (which, at 21°15' South, is almost 350 miles south of the latitude of Ofu) and to be in a better position to catch any passing troughs.

For the next two days, July 8 and 9, the trades continued blowing from between east by south and east-south-east. As long as the wind had a major easterly component, Hōkūle'a was able to hold a course to the south by sailing on the port tack. Once, however, when the wind shifted to south-east, the canoe was put on to the starboard tack to avoid being forced to the west and briefly headed east-north-east until the wind shifted round to the east.

On July 10 the wind had begun to shift to slightly north of east as Hōkūle'a reached the axis of the high pressure ridge and a low pressure trough approached from the south-west (Figure 10). Nainoa correctly read this counterclockwise shift of the winds as a sign that a trough was approaching, and that the wind would progressively shift counterclockwise around the compass as the trough reached, and then passed, the canoe on its eastward passage across the ocean. Nainoa, therefore, put the canoe on a south-east course, and made ready to sail due east before the westerly winds that the passage of the anticipated trough would bring. The next day (July 11), the wind continued shifting to the north as expected, and then became light and at times variable, a sure sign that the trades were breaking down. Once the wind was coming from the north, Hōkūle'a was able to sail due east in fairly

- 276
Surface weather map for July 10, 1986.

clear skies and calm seas. On July 12th, a line of dark clouds to the south-west announced the immediate approach of the trough. The crew tried to keep the canoe sailing in front of the trough, but it was moving much faster east than the canoe could sail, and it soon engulfed the canoe with clouds and rough seas. As the trough passed, the wind shifted quickly to the north-west, then the west, and then south-west (Figure 11), all directions which allowed the canoe to continue to be sailed due east. By the morning of the 13th, however, the trades started coming back as the wind shifted from south to south-south-east and then, in the late afternoon, started blowing from the south-east, which forced the canoe to sail to the east-north-east.

- 277
Surface weather map for July 12, 1986.

On July 14 and 15 the boisterous trades had more of an easterly component, forcing the canoe on to a north-easterly course. Nainoa figured that, if they continued sailing north-east for a day or so more, they could then tack south to Rarotonga. At this point, the Dorcas, the yacht serving as our escort vessel, began experiencing trouble in the strong winds and heavy seas. Hōkūle'a, which was rigged with extra large sails in anticipation of having to tack in light airs, was moving too fast for the heavy cruising yacht, and, at one point in squally weather, was forced to lower all the sails to wait for the Dorcas. Then, on the evening of the 15th, the Dorcas signalled that she was having trouble with her mainmast and would have to head for a port to effect repairs. - 278 Accordingly, by mutual agreement, Dan Wright, the captain of the Dorcas, disclosed to Nainoa, who was navigating without instruments, the canoe's position so that an accurate course could be laid to the nearest island, Aitutaki, which lies 140 miles due north of Rarotonga. Nainoa then tacked south to Aitutaki, arriving there on the morning of July 16.

As on previous voyages, Nainoa had been keeping track of the canoe's progress by his method of dead reckoning which is based on visual estimates of speed and course made good, as well as his estimates of the effect of the current (Finney et al. 1986:56). At the time Dan Wright notified him of the canoe's position, Nainoa found that he had underestimated the eastward progress of the canoe by about 125 miles. This margin of error, which is much larger than usual for Nainoa, may have been primarily due to the roundabout course sailed, and the consequent difficulty of estimating the canoe's progress, as well as the direction and strength of the current. In addition, because Hōkūle'a was sailing with a set of sails larger than the ones usually employed, Nainoa may have slightly, but consistently, underestimated the canoe's speed.

Aitutaki lies almost 650 miles to the east-south-east of Ofu. Hōkūle'a had reached there in a little over eight and a half days, sailing a roundabout course dictated by the direction of the winds and the strategy Nainoa evolved for making maximum easting. Although the passage of the low-pressure trough brought winds favourable for sailing to the east for only a little over three days, the easting gained then, plus the smaller amount gained when forced to sail to the north-east with the return of the trades after the passage of the trough,

Sailing track (based on satellite navigation fixes taken from a following yacht) of Hōkūle'a from Samoa to Aitutaki during July, 1986, and winds (indicated by long, straight arrows) encountered along the way.
- 279
Surface weather map for July 16, 1986.

was enough to enable the canoe to reach Aitutaki, from where Rarotonga could easily be reached once the canoe was ready to sail again (Figure 12).

Had Hōkūle'a not stopped at Aitutaki, and instead had been kept sailing with the trades on the same north-east course, the canoe would have moved into a position from which Tahiti could have been reached. On the afternoon of July 16, by which time the canoe was anchored at Aitutaki, a depression moving east at 25° South began spreading westerlies northward, and the wind in the Cook Islands began shifting to the north as the trough approached (Figure 13). Had the canoe stayed at sea, it would probably have been possible - 280 to work far enough to the east by exploiting the winds of this passing trough either to reach Tahiti direct, or at least to be in a position to reach there by tacking against the trades and then catching the westerlies accompanying the passage of the next trough.


The crew reassembled at Aitutaki in early August to make the canoe ready to sail first to Rarotonga, and from there on to Tahiti. As they were working on the canoe yet another low-pressure system transited the seas to the south, with its associated trough reaching far north into the tropics. By the time the

Surface weather map for August 10, 1986.
- 281

canoe was ready, on August 10, the trough had passed and clear weather with easterly trade winds had been re-established (Figure 14). These were ideal for sailing directly south to Rarotonga, and the canoe left just before noon. Chad Babayan, who had been practising noninstrument navigation under Nainoa's tutelage on previous voyages, was the navigator. While Aitutaki was still in sight, Babayan backsighted on a small motu (islet) on the fringing reef, keeping it aligned with one of the peaks of the central mountain core of the island in order to maintain a course due south. As the motu was lost from view, Babayan switched to steering by the descending sun. During that night he used the stars, particularly those in the Southern Cross, to keep the canoe on course, and was rewarded with the predawn sighting of Rarotonga dead ahead.


Nainoa intended to employ the same strategy to reach Tahiti as had been used on the Samoa to Aitutaki leg, that of exploiting favourable winds brought on by the passage of low-pressure troughs. As Rarotonga lies over four degrees south of Tahiti (latitude 17°30' South), the canoe was in a good position to reach Tahiti through a combination of sailing east with the westerlies, and then north-north-east against the trades whenever they reasserted their dominance. Accordingly, Nainoa planned to wait for the passage of a trough in order to sail eastwards, keeping south of Tahiti until forced to the north by the return of the trades.

The canoe entered Avarua Harbour on Rarotonga in the early afternoon of August 11. As the wind had already begun to swing to the north, indicating the approach of another trough, it was tempting to leave that evening to take full advantage of the winds, but a feast and an exchange of speeches and gifts between the ariki (traditional chiefs) of Rarotonga and the leaders of the Polynesian Voyaging Society awaited ashore.

The canoe left on the following morning, sailing east on the port tack with 12-knot northerly winds (Figure 15). 3 By the morning of the next day, August 13, the wind had shifted to the west. Around 9:00 a.m. the canoe was engulfed by heavy rain, and the wind appeared to switch from north-west to south-west, apparently indicating that the trough was passing (Figure 16). The rainy, overcast skies, and the seas confused by shifting winds, made navigation and accurate steering most difficult throughout the day, for neither sights of the sun nor observation of regular swells could be used. In fact, Nainoa was tempted at times to take the sails down and drift until he was able to gain a firm indication of direction. As, however, navigational accuracy was less important than making easting, Nainoa kept the sails up and tried to keep track of

- 282
Surface weather map for August 12, 1986.

all the wind shifts and maintain the canoe on an easterly heading. To Nainoa, the wind appeared to switch to the south at midday, and then, at sunset, to the south-east, as would be expected with the passage of a trough.

That night, however, the wind started shifting back, clockwise, towards the south-west, and then went calm. Nainoa had the sails triced (i.e., booms and sails drawn up to the mast) while waiting for the wind to pick up again. It was a most confusing night. Judging from the few directional cues he could get from occasional glimpses of the moon through the overcast, and from studying the turbulent seas, Nainoa reckoned that, when the wind did start to pick up again, it had shifted further round in a clockwise direction to the north.

- 283
Surface weather map for August 13, 1986.

This did not immediately make sense, for, had the trough passed, the wind should have continued to shift counterclockwise until it came from the east-south-east trade-wind direction, as had occurred on the way to Aitutaki. So, Nainoa kept the sails triced for a few hours until he finally got a glimpse of Jupiter and the Moon setting which confirmed that the wind had indeed shifted back to the north, and indicated that, somehow, they had moved foreward of the trough again. He then ordered the sails let out and the canoe be put on the port tack to resume sailing eastwards.

Judging from the meteorological charts consulted after the voyage, a small depression travelling in the rear of the trough along about 27° South had

- 284
Surface weather map for August 14, 1986.

generated a strong north to north-west wind field which merged with and reactivated the original trough (Figure 17).

Although sailing so close to the trough brought favourable winds, the overcast skies associated with the reactivated trough (Figure 18) continued to make navigation extremely difficult. On the night of the 15th, Nainoa made an observation of a star in the constellation Cassiopeia, the only good star sighting he had been able to make for several nights. According to his method of estimating latitude from the elevation of a star at the time of its meridional passage (see Finney et al. 1986:56–8, 65–6), this sighting indicated that the canoe was sailing along latitude 23° South, not along 20°25' as Nainoa had

- 285
Surface weather map for August 16, 1986.

mentally calculated by his dead reckoning system (based on a cumulative estimate of course and distance made good; see Finney et al. 1986:56). If true, that would mean they might be getting uncomfortably close to the Austral Islands and, therefore, should turn more towards the north. On the next day, August 16, the sight of patches of seaweed drifting from north to south confused the situation further. If, on the one hand, they had been sailing as far south as 23°, then they could already be in the midst of the Austral Islands, and the seaweed might be coming from Rurutu, the northernmost island of that group. If, on the other hand, they were at 20°25', the seaweed would most likely to be coming from the westernmost islands of the Society chain far to - 286 the north of them.

This confusion over the canoe's position was cleared up that night when several star observations indicated to Nainoa that his observation of the previous night was in error, and that they were really sailing at about latitude 21°, safely north of the Australs, but still well below the Societies. (Satellite navigation fixes from the escort yacht Dorcas indicate that Hōkūle'a was actually sailing at about 20°15' at this time, some 45 miles north of where Nainoa thought.) So sure, in fact, was Nainoa that they were not within close range of land that he ignored the sighting made that day of a few boobies and white terns, land-based birds which can mean that land is nearby. Nainoa had

Surface weather map for August 17, 1986.
- 287

learned through numerous experiences at sea that the sight of a few land birds was not a conclusive sign that an island was close. Only if he saw a lot of birds, or groups flying in the same direction together (particularly at sunrise when they could be coming out from their island to their fishing grounds, or at sunset when they could be returning home) was Nainoa willing to accept that land must be near.

At this point, Nainoa would have liked to have been closer to Ra'iatea, for plans called for a stop there to pay a ceremonial visit to the great marae of Taputaputea before making landfall on Tahiti. However, the northerly winds would not allow the canoe to be sailed due north to Ra'iatea. With northerly winds, the canoe could continue to sail on the port tack to the east-north-east, or go over on to the starboard tack and sail to the west-north-west. As Nainoa did not want to tack back towards the west, he gave up the idea of visiting Ra'iatea and concentrated on trying to make direct for Tahiti.

By dawn on the next day, August 17, it began to look like it would be difficult even to make Tahiti. If the northerly winds persisted, Hōkūle'a would be forced to sail right past Tahiti well south of the island. For a while that day the canoe did seem to get out in front of the trough, and move into a region of clearing skies and east-north-east winds which dried the deck out for the first time on the voyage (Figure 19). Nainoa took the wind shift and dry air to indicate that the canoe was entering into a ridge of high pressure, and that the trade winds might soon come back in force. The east-north-east winds did allow the canoe to be sailed almost directly to the north on the starboard tack, but during that night the wind shifted back to the north as the trough gathered speed and began to catch up with Hōkūle'a, forcing the canoe back on the port tack and on to an easterly heading. As the squally winds continued to blow from the north, Nainoa realised that not only were they going to sail past Tahiti, but also that they might even be blown into the hazardous labyrinth of atolls of the Tuamotu Archipelago. After worrying for months how they were going to make enough easting to reach Tahiti, Nainoa realised that the problem might soon be how to make enough westing to sail back to the island!

As the north-west winds kept forcing the canoe to the east, the crew could look back to the south-west where a band of towering cumulus clouds appeared to mark the front of the trough. If only, they thought, the canoe could somehow get to the rear of the trough they would find trade winds that would allow them to sail north-west back to Tahiti. Finally, at sunset on the 19th, the wind started to shift to the north-east, indicating perhaps that the trough was about to pass them. Then the wind died altogether, and, after a brief calm, a succession of squalls moved in from the west. The squalls, accompanied by winds blowing up to 40 knots, persisted throughout the night, periodically

- 288
FIGURE 20. Surface weather map for August 21, 1986.

forcing the sails to be triced and lowered to the deck. At dawn, a strong wind and a big swell started coming from the south-south-west, indicating that the trough must be passing, and that they were almost back in the trades. Accordingly, Nainoa ordered the canoe to be sailed to the north-north-west, the direction he thought might bring them to Tahiti.

Although Nainoa was sure that Tahiti was somewhere to the north and west of them, he was not sure exactly where the canoe was in relation to Tahiti because of the cloud cover that had plagued them throughout the voyage and which had been especially heavy during the last few nights. Accordingly, he planned to keep sailing to the north-north-west until sunset, when, if they had - 289 not spotted Tahiti, he intended to start tacking back and forth to the south-west to search out the island.

The latter part of that strategy never had to be put into action. After a few hours of sailing, the mist parted to reveal, just a few miles ahead of the canoe, the precipitous volcanic peak of Mehetia, a tiny island lying 60 miles east-south-east of Tahiti. From then on, the sail to Tahiti was straightforward. The canoe passed Mehetia, and, as its lone peak began to disappear astern, Tahiti could be seen rising out of the sea ahead. By then the trough lay to the east of the canoe's position, the skies had cleared, and the winds had continued their counterclockwise rotation so that Hōkūle'a was sailing in strong south-east trades for the first time during this leg (Figure 20).

By the late afternoon the canoe was sailing along the rugged shore of Tahiti-iti, the uninhabited eastern end of the island which is visited only rarely by fishermen and sightseers. The sight of the verdant mountain slopes unscarred by roads or cultivation, and of secluded little bays without a house or other sign of human presence, made the crew feel that they were sharing a landfall experience with the first discoverers of this island. After waiting offshore that night, the canoe was sailed into the pass at Tautira, a village located midway along the northern shore of the Tahiti-iti peninsula. There a welcome from the mayor, Tutaha Salmon, and the many friends of Hōkūle'a and her crew from the Maire-Nui canoe club, awaited the weary voyagers.

The voyage from Rarotonga to Tahiti had been at once unforeseeably easy and unexpectedly difficult. Nainoa had anticipated a hard struggle to get the required easting to reach Tahiti — by exploiting the westerly winds of whatever passing troughs they might encounter, and by tacking against the trade winds when necessary. Instead, an embarrassment of northerly and westerly winds sped the canoe eastwards as she stayed within or adjacent to a low-pressure trough for virtually the entire distance. Even though the canoe was forced to sail past Tahiti, and then, when the trades finally returned, to sail back to the north-west, Hōkūle'a still covered the 720 miles between Tahiti and Rarotonga in just eight and a half days.

Ironically, although the trough generated the winds needed for sailing east, it also brought the cloudy, squally weather that Nainoa had wanted to avoid by sailing during the winter instead of the summer. The frequent squalls forced the crew to trice the sails and lower them to the deck so many times that they lost count, and made the job of navigating without instruments most difficult. Even so, although Nainoa did not know exactly where the canoe was when the trades finally returned and allowed the canoe to be tacked back to the north-west, his notion of the canoe's general location in relation to Tahiti passed the test of landfall — first on Mehetia and then on Tahiti itself.

- 290
FIGURE 21. Sailing track (based on satellite navigation fixes taken from a following yacht) of Hōkūle'a from Aitutaki to Tahiti during August, 1986, and winds (indicated by long, straight arrows) encountered along the way.

If any experience could lay to rest the myth of the permanence of the trades in Polynesian waters, it would be this one. During the Austral winter, when the trades are supposed to be steadiest, Hōkūle'a had sailed west to east from Rarotonga to Tahiti using northerly and westerly winds that drove the canoe past the intended landfall before the trades finally returned and allowed Hōkūle'a to be sailed back to the north-west to reach Tahiti (Figure 21).


The favourable winds that enabled Hōkūle'a to reach Tahiti may be labelled “subtropical westerlies” in that they originated from troughs extending north from subtropical depressions, not, as in the case of the regular monsoonal westerlies, from the influence of the monsoon trough generated by the summer heating of northern Australia. (Even though the passing troughs may generate spells of northerly as well as westerly winds, as was demonstrated during the voyage, we prefer the simpler label “subtropical westerlies” to the more cumbersome one of “subtropical northerlies and westerlies”.) Although these subtropical westerlies may be much less frequent and enduring than the monsoonal westerlies, they none the less seem to develop periodically. In a meteorological study of the Southern Hemisphere, Trenberth and Mo (1985) found that the region south-east of the Samoa-to-Tahiti route is prone to episodes when high-pressure systems block the normal passage of lows, shunting some of them on to a more northerly course. On the basis of weather records from Rarotonga, and the testimony of experienced sailors and fishermen there, we can roughly estimate that frequent and - 291 prolonged outbreaks of subtropical westerlies, such as those encountered sailing from Samoa to Tahiti, may occur in Rarotongan waters in about one year out of 10. To the monsoonal westerlies of the summer, we must therefore add the occasional outbreak of subtropical westerlies as a wind regime that pioneering voyagers could have exploited to expand eastwards.


Our voyage has demonstrated how it is possible to sail a voyaging canoe from west to east across Polynesia by using westerly wind shifts, and that it can be done during the Austral winter as well as during the Austral summer. In addition to the issue of how the migration from West to central East Polynesia was accomplished, our experience relates directly to the questions of when this was accomplished, which islands may have been involved, and to what degree there could have been two-way communication within central East Polynesia and between there and West Polynesia, which are so crucial to our understanding of the development of East Polynesian cultures (Kirch 1986).

The earliest sites found so far in East Polynesia proper are in the Marquesas Islands, and have been dated, though not without controversy, to around the second century B.C. (Suggs 1961; Kirch 1984:73; 1986:27; Ottino 1985; Sinoto 1970). 4 However, it appears that Fiji, Tonga and Samoa may have been occupied as early as 1600 B.C. by voyagers bearing the distinctive Lapita Cultural Complex from which Polynesian culture springs (Kirch and Hunt, in press). Furthermore, the archaeological evidence shows that the movement of these Lapita voyagers from the Bismarck archipelago off the north-east coast of New Guinea to these central Pacific islands was so fast as to be “practically instantaneous” (Irwin 1980:325; 1981:483) in the sense of being “undetectable within the statistical uncertainty range of radiocarbon dating and of archaeological sampling error” (Kirch and Hunt 1988). If we take this to mean two or three centuries, why would it have taken a dozen or so centuries more for the descendants of these voyagers to have reached the Marquesas, lying some 1800 nautical miles from Samoa, roughly equivalent to the distance from New Britain in the Bismarck Archipelago to Fiji? 5

At least until recently, the most commonly considered hypothesis was that there was a significant pause in migration once the Lapita voyagers reached Fiji, Tonga and Samoa, and that, only after a millennium or more of local adaptation and change within these rich and varied archipelagos resulting in the development of the ancestral Polynesian culture pattern, did the eastward migration resume (Bellwood 1978a:54-9, 1979:318; Finney 1979:343; Green - 292 MS 1987b). Now, however, archaeologists are paying close attention to Irwin's (1981:486) proposal that there was no long pause in eastward migration, and that sites earlier than those in the Marquesas should be found in the East Polynesian islands closest to Samoa and Tonga, where evidence of early human settlement has so far escaped the attention of archaeologists either because it has been destroyed by many centuries of gardening activity or wave action, or because it has been deeply buried by eroded soil or even submerged beneath the surface of the sea.

The demonstration by Spriggs (1985) and others how extensively soil washed down from hillsides cleared and burned for agriculture can alter the landscape along the coast of an island, and advances in oceanic geology indicating that Samoa and the islands immediately to the east are subsiding relatively rapidly as that portion of the westward-moving Pacific Plate is being subducted under the Fijian Plate (cited in Kirch and Hunt, in press), are stimulating archaeologists to address the question of the continuous or discontinuous nature of the migration eastwards across the Pacific by developing excavation methods appropriate to searching for deeply buried or submerged sites on islands immediately to the east of West Polynesia. Although these searches are only now getting under way, discoveries from the atoll of Pukapuka in the Northern Cooks (Chikamori MS 1987), and the high island of Atiu in the Southern Cooks, 6 would already seem to indicate that substantial evidence of early migration may be found in this region.

From our perspective of sailors who have tried to gain insight into Polynesian migrations by sailing a canoe over the routes in question, it would not be surprising if the movement from West Polynesia across central East Polynesia as far east as the Marquesas did turn out to have taken somewhat longer than that from the Bismarcks to Fiji, Tonga and Samoa. The wind regime along the Lapita migration trail in the south-western Pacific, where westerlies dominate during the Austral summer, would seem ideal for exploring and colonising rapidly to the east, whereas the more intermittent character of the westerlies along the Polynesian migration trail in the south-eastern Pacific would seem less condusive to rapid movement. It would seem unlikely, however, that a full thousand years or more would have been needed for sailors who had reached the edge of Polynesia to learn how to exploit the shorter-lived and less frequent spells of westerlies there in order to colonise the archipelagos lying to the east. The archaeological record from Fiji, Tonga and Samoa indicates that there was extensive voyaging among these archipelagos (Davidson 1976:45; Green 1975). Coping with the wind conditions in this mid-Pacific region where the westerlies are more intermittent than in the Lapita homeland region could have prepared voyagers for the more - 293 difficult conditions to the east. Kirch (1984: 52) estimates that ancestral Polynesian technology is archaeologically distinguishable from its Lapita roots by about 500 B.C., by which time voyagers should have been well acquainted with, and adapted to, the wind regime in this part of the Pacific. It would, therefore, seem likely that archaeologists should be able to find evidence of voyagers reaching East Polynesian islands at least as early as 500 B.C., if not several centuries before then (cf. Irwin 1981:484, 489).

Although it would have been possible to sail a double-canoe direct from West Polynesia to the Marquesas (Finney 1985: 16-18) during a massive outbreak of westerlies, such as occurred during the 1982-3 El Nio, there is nothing about the wind patterns or the sailing characteristics of double-canoes that would necessarily have channelled all eastward initiatives that way. All the central East Polynesian archipelagos could have been reached direct from West Polynesia. Both the Northern Cooks and the Southern Cooks are within reasonable sailing distance of Samoa — as witness the ethnographic information to be cited below of voyaging between Pukapuka and Samoa, as well as the record of our voyage to Aitutaki and Rarotonga. Because of their more southerly position, Tongan sailors might have been in an even better position than those in Samoa to exploit low-pressure troughs extending up from temperate latitudes, and the accompanying subtropical westerlies. Some voyagers heading east from Tonga or Samoa might even have been forced by northerly winds — such as we experienced on the second leg of our voyage — all the way to the Austral Islands. Nor can the possibility of the early settlement of the Societies direct from West Polynesia be ignored. Had there been no obligation to return to the Cooks, Hōkūle'a could probably have continued sailing eastwards, exploiting episodes of subtropical westerlies until the Societies were reached. Archaeologists working in central East Polynesia should be prepared to find evidence of direct settlement from the west on several archipelagos, not just one.

In addition to reshaping our thinking to encompass multiple landfalls by colonists throughout the central archipelagos of East Polynesia, we should also be alert to the probability that the early colonies were in regular, or semi-regular contact with one another. Fifteen years ago, Biggs (1972: 149) warned against “any simplistic view of Polynesian settlement passing from A to B to C in a sequence which never retraces its steps. . .”, and proposed that, as more detailed research is completed in various fields, a more realistic “theory of intra-Polynesian migration and settlement” will emerge. Our experience supports Biggs' reasoning. We can conceive of no technical reason why, once the various islands and archipelagos were settled, there would have been no contact among them.

- 294

Yet, when Europeans first sailed into Polynesian waters, interarchipelago voyaging in East Polynesia seems to have been largely restricted to such nearly contiguous groups as the Societies and the north-west Tuamotu atolls. Although it might seem logical to assume that, if Polynesians were not voyaging extensively between archipelagos at the time of European contact, they had never done so in the past, it should be recognised that such reasoning is based on a modern Western assumption of unilineal progress which may not be applicable to all places and ages. If anything, the evidence from throughout the Pacific suggests that, after rapid expansion into island regions, and a subsequent period of wide-ranging interisland communication, voyaging spheres contract as individual island societies grow and mature (cf. Irwin 1980:328). This expansion-contraction cycle is beginning to be well documented in the archaeological and ethnological record along the Lapita migration trail (Allen 1984:429-32; Green 1974:253; 1987a). For East Polynesia, a cycle involving a period of two-way voyages and then a cutting of ties is reflected in some oral traditions. For example, those from Hawai'i tell of a period of voyaging back and forth between there and Tahiti which continued over several generations before ceasing (Fornander 1916, 1919; Emerson 1893), while traditions from both Ra'iatea and Rarotonga independently tell of voyages that were once regularly made between these islands (Henry 1928:121-7; Williams 1837:47-8, 88), and how conflict ended regular communication. 7

Although, in recent decades, it has been popular to deny that Polynesian voyaging legends relate to actual events and situations, our experience in sailing Hōkūle'a round Polynesia indicates that it is feasible to sail a double-canoe along the routes specified in the oral traditions, and that, at least in the case of the voyage from Rarotonga to Aotearoa (New Zealand) (Babayan et al. 1987), these traditions contain vital information on course headings and the right time of the year to sail. While the only round-trip between widely separated islands within central East Polynesia that we have made was between Tahiti and Rarotonga, we believe that we could have sailed Hōkūle'a among all the central archipelagos of East Polynesia. 8 Surely, earlier, and probably more skilled, voyagers could have sailed their canoes over these routes in the past.

If there was such interarchipelago contact early in the settlement period, when populations would still have been small, it would seem likely that there would also have been significant linguistic and cultural interchange among the newly founded colonies. The search for a single East Polynesian homeland whence all other East Polynesian cultures sprang may, therefore, prove as elusive as was a similar attempt in West Polynesia to pinpoint either Tonga - 295 or Samoa as the single fount of Polynesian culture (Davidson 1976; Green 1981). Building on the work of Hunt (MS 1979), Green (1981, 1985) and others, Kirch (1986) has argued in his recent rethinking of the prehistory of East Polynesia that the first stages of cultural development there may have proceeded within a multi-archipelago region of communicating colonies, not on an individual island or archipelago.

Nor can we rule out the possibility of some voyaging back and forth between islands in East Polynesia, such as those in the Cooks, and islands in West Polynesia after the initial colonisation voyages (and whatever two-way voyaging that might have involved). The most solid evidence of any such two-way communication comes from the atoll of Pukapuka in the Northern Cooks, which is not surprising, for the island lies less than 300 miles east-north-east of the Manu'a Group of Samoa, and is intermediate in culture between West and East Polynesia (Burrows 1938; Vayda 1959). Over 50 years ago, the writer Robert Dean Frisbie recorded and analysed sailing directions between Pukapuka and various islands to the west (Beaglehole and Beaglehole 1938:351-3). Although these appeared impossibly precise to such students of Polynesian navigation as Åkerblom (1968:26) and Sharp (1964:89), the star courses and seasonal choice of departure times do make good navigational and sailing sense from what we have learned in sailing Hōkūle'a around Polynesia. For example, to sail to Olosenga, the island in the Manu'a Group next to Ofu, the Pukapukans used easterly trade winds and followed a star course calculated to bring them to this tiny Samoan island lying to the south-south-west. To return home, they waited for the onset of the monsoon westerlies, sailing eastwards on a reciprocal star course as early as possible in the season in order to avoid, Frisbie hypothesised, the hurricanes that are more prevalent later in the summer.

For the more distant Southern Cooks, or any other East Polynesian archipelago, we have no definite ethnographic or historical information of two-way voyaging to and from West Polynesia. None the less, both oral traditions and an unusual archaeological find from Rarotonga would seem to point to some voyaging relations between the Southern Cooks and Samoa. Rarotongan legends celebrate Karika, a chiefly voyager supposedly from the Manu'a group in Samoa who settled in Rarotonga around A.D.1300 (Williams 1837), and whose ariki descendants welcomed Hōkūle'a and her crew to that island. A cache of adzes with Samoan affinities found buried within an ancient dwelling site could be interpreted either as evidence of a Samoan intrusion at about the same time as Karika (Bellwood 1978b:201), or as Cook Island influence in Samoa (cf. Green and Davidson 1974:261), or perhaps of two-way communication. 9

- 296

One of the outstanding discoveries concerning the Lapita migration is that of extensive two-way voyaging over great distances, as indicated by the transport of obsidian and other valuable materials (Ambrose and Green 1972; Green 1974, 1987a; Best 1987). The regular alternation of a full season of westerlies with the easterly trade winds in the south-western Pacific must have facilitated two-way voyaging along the Lapita end of the voyaging trail. It will be interesting to see if more intensive archaeological research within Polynesia indicates if early Polynesian voyagers were able to cope with the greater dominance of the trades and the more episodic nature of the westerlies in their part of the Pacific in order to be able to sail back and forth with any frequency between East and West Polynesia.


Without the ability to sail over long distances, to find islands strewn over many thousands of miles of open sea, and to carry enough people, tools, plants and animals to found viable colonies on the islands discovered, there would have been no Polynesian culture, no vast triangular section of ocean occupied by closely related neolithic peoples. That the large, stable, and seaworthy double-canoe was the critical artefact of this cultural development and expansion is generally accepted, just as the ability to make one's way across the ocean and find distant islands by reading the stars, the winds, the swells, the flight of birds, and other clues provided by nature is often cited as the skill most crucial to this process. To the double-canoes, and ways of navigating them without instruments, we would add a third main element of this oceanic adaptation that made the colonisation of so many far-flung islands possible: knowledge of the winds of the sea and the skill to exploit spells of westerly winds to sail far to the east.

While the more intermittent character of the westerlies in the tropical south-eastern Pacific may have slowed the momentum of eastward expansion across the Pacific, the ethnographic and experimental evidence suggests that early Polynesian voyagers were able to adapt to this wind regime and to use periodic episodes of westerly winds to find and settle all the oceanic islands to the east of their mid-Pacific homeland. The evidence further suggests that they would have been able to make the multiple landfalls throughout central East Polynesia, and that, once settled on the various islands and archipelagos, they and their descendants would have been capable of exploiting the alternating rhythm of monsoonal and subtropical westerlies with easterly trade winds to maintain some communication ties within the central East Polynesia region, and also to some extent between East and West Polynesia. The actual history of East Polynesia colonisation may, therefore, turn out to - 297 be much more complex than suggested by broad arrows commonly drawn on maps to indicate migration paths. Instead of searching for a single island or archipelago as the sole site of first settlement in East Polynesia, and of assuming one-way population dispersal from there to all the other eastern archipelagos, perhaps we should think of early East Polynesian colonisation in terms of a large multi-archipelago, intercommunicating region with some two-way links back and forth between there and West Polynesia. Although the camp-sites and settlements made by the first people to reach the islands of East Polynesia may be sparse and difficult to find, and evidence of interisland communication even harder to discern, the archaeologists should keep looking for evidence of early colonies and their interrelations throughout the islands and archipelagos lying to the east of the Polynesian homeland. 10

- 298
  • Åkerblom, Kjell, 1968. Astronomy and Navigation in Polynesia and Micronesia. Stockholm, Ethnografiska Museet Monograph Series, No. 14.
  • Allen, Jim, 1984. Pots and Poor Princes: A Multidimensional Approach to the Role of Pottery Trading in Coastal Papua, in S. E. van der Leeuw and A. C. Pritchard (eds),
- 299
  • The Many Dimensions of Pottery: Ceramics in Archaeology and Anthropology. Amsterdam, Institute for Pre- and Protohistory, University of Amsterdam. pp.409-73.
  • Ambrose, Wallace R. and Roger C. Green, 1972. First Millennium BC Transport of Obsidian from New Britain to the Solomon Islands. Nature, 237:31.
  • Babayan, Chad, Ben Finney, Bernard Kilonsky and Nainoa Thompson, 1987. Voyage to Aotearoa. Journal of the Polynesian Society, 96:161-200.
  • Banks, Joseph, 1962. The Endeavour Journal of Joseph Banks 1768-1771. J. C. Beaglehole, (ed.). 2 Vols. Sydney, Angus and Robertson.
  • Beaglehole, Ernest and Pearl Beaglehole, 1938. Ethnology of Pukapuka. Honolulu, B. P. Bishop Museum Bulletin 150.
  • Beechey, F. W., 1831. Narrative of a Voyage to the Pacific and Beering's Strait, to Co-operate with Polar Expeditions. 2 Vols. London, Henry Colburn and Richard Bentley.
  • Bellwood, Peter, 1978a. The Polynesians. London, Thames and Hudson.
  • —— 1978b. Archaeological Research in the Cook Islands. Pacific Anthropological Records 27.
  • —— 1979. Man's Conquest of the Pacific. New York, Oxford University Press.
  • Best, Simon, 1987. Long Distance Obsidian Travel and Possible Implications for the Settlement of Fiji. Archaeology in Oceania, 22:31-2.
  • Biggs, Bruce G., 1972. Implications of Linguistic Subgrouping with Special Reference to Polynesia. Pacific Anthropological Records, 3:143-52.
  • Burrows, Edwin G., 1938. Western Polynesia, a Study in Cultural Differentiation. Ethnologiska Studier, 7:1-92.
  • Chikamori, Masashi, MS 1987. East and West in the Samoic group: Archaeology on Rennell and Pukapuka. Abstract of a paper presented at the International Congress on Development and Isolation in the Pacific, Osaka-Tokyo, September 1-7.
  • Cook, James, 1955. The Journals of Captain James Cook on his Voyages of Discovery: The Voyage of the Endeavour, 1768-1771. J. C. Beaglehole (ed.). Cambridge, Hakluyt Society.
  • Corney, Bolton G., 1915. The Quest and Occupation of Tahiti by the Emissaries of Spain in 1772-1776. Vol. 2. London, Hakluyt Society.
  • Davidson, Janet, 1976. The Prehistory of Western Polynesia, in José Garanger (ed.), La Préhistoire Océanienne. Nice, IXe Congrés, Union Internationale des Sciences Préhistoriques et Protohistoriques. pp.27-51.
  • Dickenson, William R., and Richard Shutler, Jr., 1974. Probable Fijian Origin of Quartzose Temper Sands in Prehistoric Pottery from Tonga and the Marquesas. Science, 185:454-7.
  • Emerson, Nathaniel B., 1893. The Long Voyages of the Ancient Hawaiians. Papers of the Hawaiian Historical Society, No. 5.
  • Emory, Kenneth P., 1959. Origin of the Hawaiians. Journal of the Polynesian Society, 68:29-35.
  • Feinberg, Richard, 1986. On “Anomalous Westerlies, El Niño, and the Colonization of Polynesia” by Ben R. Finney. American Anthropologist, 88:454-5.
- 300
  • Finney, Ben R., 1977. Voyaging Canoes and the Settlement of Polynesia. Science, 106:1277-85.
  • —— 1979. Voyaging, in Jesse Jennings (ed.), The Prehistory of Polynesia. Cambridge, Harvard University Press. pp.323-51.
  • —— 1985. Anomalous Westerlies, El Niño, and the Colonization of Polynesia. American Anthropologist, 87:9-26.
  • —— n.d. Polynesian Voyagers to the New World. (A manuscript to be published in a forthcoming festschrift.)
  • —— Bernard Kilonsky, Steven Somsen, and Edward D. Stroup, 1986. Reviving a Vanishing Art. Journal of the Polynesian Society, 95:41-90.
  • Firth, Raymond, 1936. We, the Tikopia. London, George Allen and Unwin.
  • Fornander, Abraham, 1916. Fornander Collection of Hawaiian Antiquities and Folklore. Memoirs of the Bernice Pauahi Bishop Museum, 4(1):112-73.
  • —— 1919. Fornander Collection of Hawaiian Antiquities and Folklore. Memoirs of the Bernice Pauahi Bishop Museum, 6(2):239-57.
  • Green, Roger C., 1967. The Immediate Origin of the Polynesians, in G. A. Highland et al. (eds), Polynesian Culture History. Honolulu, B. P. Bishop Museum. pp.215-40.
  • —— 1974. Sites with Lapita Pottery: Importing and Voyaging. Mankind, 9:253-9.
  • —— 1975. Polynesian Voyaging. Science, 187:274.
  • —— 1978. New sites with Lapita pottery and their Implications for an Understanding of the Settlement of the Western Pacific. Working Papers in Anthropology, Archaeology, Linguistics and Maori Studies No. 51. Auckland, Department of Anthropology, University of Auckland.
  • —— 1979. Lapita, in Jesse D. Jennings (ed.), The Prehistory of Polynesia. Cambridge, Harvard University Press. pp.27-60.
  • —— 1981. Location of the Polynesian Homeland: A Continuing Problem, in Jim Hollyman and Andrew Pawley (eds), Studies in Pacific Languages and Cultures. Auckland, Linguistic Society of New Zealand. pp.133-58.
  • —— 1985. Sub-grouping of the Rapanui Language of Easter Island in Polynesian and its Implications for East Polynesian Prehistory. Working Papers in Anthropology, Archaeology, Linguistics, Maori Studies, No.65. Auckland, Anthropology Department, University of Auckland.
  • —— 1987a. Obsidian Results from the Lapita Sites of the Reef/Santa Cruz Islands, in W. R. Ambrose and J. M. J. Mummery (eds), Archaeometry: Further Australasian Studies. Canberra, Department of Prehistory, Australian National University. pp.239-49.
  • —— MS 1987b. Peopling of the Pacific: a Series of Adaptive Steps: or Punctuated Evolution. Paper presented at Section H, 57th Annual Meeting of A.N.Z.A.A.S., August 24.
  • —— and Janet M. Davidson, 1974. Archaeology in Western Samoa. Vol. 2. Bulletin No. 7. Auckland, Auckland Institute and Museum.
  • Hale, Horatio, 1846. U.S. Exploring Expedition 1838-42, Ethnography and Philology. Philadelphia, Lea and Blanchard.
- 301
  • Henry, Teuira, 1928. Ancient Tahiti. Honolulu, B. P. Bishop Museum Bulletin 48.
  • Hessell, J. W. D., 1981. Climatology of the South-west Pacific Islands. 2. Climatological Statistics, in W. R. Dale (ed.), Pacific Islands Water Resources. Wellington, South Pacific Technical Inventory 2. Department of Scientific and Industrial Research. pp.35-43.
  • Heyerdahl, Thor, 1978. Early Man and the Ocean. London, Allen and Unwin.
  • Hunt, Terry L., MS 1979. Alternative Perspectives for Early East Polynesian Prehistory. Paper presented at the New Zealand Archaeology Association Biennial Conference, Auckland.
  • —— and Patrick V. Kirch, 1988. An Archaeological Survey of the Manu'a Islands, American Samoa. Journal of the Polynesian Society, 97:153-83.
  • Irwin, Geoffrey, 1980. The Prehistory of Oceania: Colonization and Cultural Change, in Andrew Sherratt (ed.), The Encyclopedia of Archaeology. New York, Cambridge University Press. pp.324-32.
  • —— 1981. How Lapita Lost its Pots: The Question of Continuity in the Colonisation of Polynesia. Journal of the Polynesian Society, 90:481-94.
  • Kane, Herb Kawainui, 1976. Voyage, the Discovery of Hawai'i. Honolulu, Island Heritage.
  • Keegan, William F. and Jared M. Diamond, 1987. Colonization of Islands by Humans: A Biogeographical Perspective, in Michael B. Schiffer (ed.), Advances in Archaeological Method and Theory. Vol. 10. New York, Academic Press. pp.49-91.
  • Kemp, Peter, 1976. The Oxford Companion to Ships & the Sea. Oxford, Oxford University Press.
  • Kirch, Patrick V., 1984. The Evolution of Polynesian Chiefdoms. Cambridge, Cambridge University Press.
  • —— 1986. Rethinking East Polynesian Prehistory. Journal of the Polynesian Society,95:9-40.
  • —— and Terry Hunt, 1988. Radiocarbon Dates from the Mussau Islands and the Lapita Colonization of the Southwestern Pacific. Radiocarbon, 30(2).
  • —— in press. The Spatial and Temporal Boundaries of Lapita, in Patrick V. Kirch and Terry L. Hunt (eds), Lapita Archaeology: A Critical Review. Seattle, Burke Museum Special Research Report No. 5, Burke Memorial Museum.
  • La Pérouse, J. F. G. de, 1798. A Voyage Around the World in the Years 1785-88. 3 Vols. London, Johnson.
  • Lewthwaite, Gordon R., 1970. The Puzzle of Tupaia's Map. New Zealand Geographer, 26:1-18.
  • Ottino, Pierre, 1985. Un Site Ancien aux îles Marquises: l'Abri-sous roche d'Anapua, à Ua Pou. Journal de la Société des Océanistes, 41(80):33-7.
  • Revell, C. G., 1981. Climatology of the South-west Pacific Islands. 1. Circulation Systems, in W. R. Dale (ed.), Pacific Islands Water Resources. Wellington, South Pacific Technical Inventory 2. Department of Scientific and Industrial Research. pp.27-34.
  • Sadler, James C. and Bernard Kilonsky, 1983. Meteorological Events in the Central Pacific during 1983. Tropical Ocean-Atmosphere Newsletter, 21:2-3.
- 302
  • Sharp, Andrew, 1964. Ancient Voyagers in Polynesia. Berkeley, University of California Press.
  • Sinoto, Yoshihiko, 1970. An Archaeologically Based Assessment of the Marquesas Islands as a Dispersal Center in East Polynesia. Pacific Anthropological Records, 11:105-32.
  • Skelton, R. A., 1955. The Journals of Captain James Cook on his Voyages of Discovery: Charts and Views drawn by Cook and his Officers. Cambridge, Hakluyt Society.
  • Spriggs, Matthew, 1985. Prehistoric Human-induced Landscape Enhancement: Examples and Implications, in I. S. Farrington (ed.), Prehistoric Intensive Agriculture in the Tropics. BAR International Series 232. pp.410-23.
  • Stair, John B., 1895. Flotsam and Jetsam from the Great Ocean: or, Summary of Early Samoan Voyages and Settlement. Journal of the Polynesian Society, 4:99-131.
  • Steiner, J. T., 1980. The Climate of the South-West Pacific Region. Wellington, New Zealand Meteorological Service Miscellaneous Publication No. 166.
  • Suggs, Robert C., 1961. Archaeology of Nuku Hiva, Marquesas Islands, French Polynesia. Anthropological Papers of the American Museum of Natural History, 49 (1).
  • Thompson, C. S., 1986a. The Climate and Weather of the Southern Cook Islands. Wellington, New Zealand Meteorological Service Miscellaneous Publication No. 188 (2).
  • —— 1986b. The Climate and Weather of the Northern Cook Islands. Wellington, New Zealand Meteorological Service Miscellaneous Publication No. 188 (3).
  • Trenberth, Kevin E. and Kingtse C. Mo, 1985. Blocking in the Southern Hemisphere. Monthly Weather Review, 113:3-21.
  • Vayda, Andrew P., 1959. Polynesian Cultural Distribution in New Perspective. American Anthropologist, 61:817-28.
  • Williams, John, 1837. A Narrative of Missionary Enterprises in the South Sea Islands. London, John Snow.
  • Wyrtki, Klaus, and Gary Myers, 1975. The Trade Wind Field over the Pacific Ocean. Part 1: The Mean Field and the Mean Annual Variation. Honolulu, Hawai'i Institute of Geophysics, University of Hawai'i.
1   In an article published after Nainoa Thompson formulated his voyaging strategy, Feinberg (1986) reports that the sailors of Anuta, a tiny Polynesian Outlier located midway between Fiji and New Guinea, do not like to chance the stormy seas of the westerly season. Instead, when they want to visit Patutaka, a neighboring island to the south-east, they sail in the trade wind season, waiting for the trades to drop and for brief spells of northerly and westerly winds that would enable them to sail on a reach or before a tail wind. Once they have reached Patutaka, they are then in a good position to return home before the wind when the trades resume blowing.
2   Milton (“Shorty”) Bertlemann served as captain, as he had on several of the previous legs of the voyage. Nainoa Thompson was the navigator. The crew included: Mau Piailug from Satawal Atoll of the Caroline Islands in the Federated States of Micronesia; Tua Pittman from Rarotonga; and Clayton Bertlemann, Bernard Kilonsky, Tava Taupu, Pauahi Ioane, Harry Ho, Melvin Paoa and Ben Lindsey from Hawai'i.
3   For this leg, Nainoa acted as both navigator and captain. Crewmen included Mau Piailug from Satawal; and Pat Aiu, Abraham (“Snake”) Ahee, Chad Babayan, Aaron Young, Bruce Blankenfeld, Glenn Oshiro, Michael Tongg, Wallace Froiseth, and Richard Rhodes from Hawai'i.
4   The uncorrected C14 ages of the culturally-associated organic materials excavated by Suggs was 2080±150 years and 1910±180 years B.P., which Kirch (1986:24) corrects for secular variation to 405 B.C.-A.D.220 and 385 B.C.-A.D.450 at 95 per cent confidence levels. Kirch rejects Sinoto's (1970) redating of the beginning of the Marquesan sequence to A.D.300.
5   The actual voyaging distance, going from island to island through Melanesia, to Fiji would, of course, be greater than the great circle distance of some 1,800 nautical miles between New Britain in the Bismarcks and Fiji. Although obsidian found in an early Fijian settlement has recently been sourced to deposits at Talasea on the Williaumez Peninsula of New Britain (Best 1987), that does not necessarily mean that Fiji was settled directly from New Britain. Similarly, although the discovery of Marquesan potsherds containing temper sands identified by Dickenson and Shutler (1974) as having come from Fiji could indicate a direct voyaging link, an island-by-island route to the Marquesas through the Cooks and perhaps also the Societies that would be significantly longer than the 1,800 nautical mile direct route is also possible.
6   A potsherd found in 1987 in secondary deposits in a site on Atiu has been analysed by William Dickenson, who is of the opinion that it contains temper sands that appears to come from a continental island source, although he could not identify that source (personal communication, Yosihiko Sinoto, May 31, 1988).
7   One possible explanation for the wide if imprecise geographical knowledge displayed in Tupa'ia's famous map (Skelton 1955: Chart XI) would be that his awareness of the general location of such distant islands as Fiji, Rotuma, and the Marquesas represented knowledge that had been passed down for generations after active long-distance voyaging had ceased. Yet, as Lewthwaite (1970:17-8) vigorously argues, Tupa'ia may himself have sailed to the Manu'a group in Samoa and then back to Tahiti. It could, therefore, be that there was more long-distance voyaging going on in contact-era Polynesia than the European explorers were aware of.
8   In April 1987, while on the way back to Hawai'i from Tahiti, Hōkūle'a was sailed first to Rangiroa Atoll to be in a position to sail to Hawai'i via the Marquesas. However, after waiting several weeks at Rangiroa for winds favourable for sailing north-north-west to the Marquesas, in order to meet the schedule, the canoe was sailed straight back to Hawai'i. Had the crew at their disposal a number of months to wait for the right winds — either westerlies or southerlies — we believe that Hōkūle'a could have been sailed to the Marquesas. Or, alternatively, had the voyage been attempted in 1986, right after the canoe arrived in Tahiti, it would have been possible to use a combination of trade winds to get clear of the Tuamotus, then a spell of westerlies to gain the needed easting to make it to the Marquesas.
9   A better picture of whatever relations the Manu'a group had with the Cook Islands and other parts of East Polynesia should emerge with more intensive archaeological work in both East Polynesia and Manu'a (Hunt and Kirch 1988). The traditions gathered from a Rarotongan by the missionary John B. Stair (1895) about the Samoan settlement of Rarotonga and wider ranging Samoan voyages should be treated with caution because of the degree to which they may include inspiration and embellishments from the early period of European contact.
10   We wish to thank the Fiji Meteorological Service, the New Zealand Meteorological Service, and the Department of Meteorology, Hawai'i Institute of Geophysics of the University of Hawai'i—and especially Thomas Schroeder of that institute—for their aid in supplying meteorological data and interpretations. We would also like to thank Roger Green, C. S. Thompson, Geoff Irwin, Jared Diamond, Terry Hunt, Doug Sutton, and Dixon Stroup for their comments on earlier drafts of this paper, as well as the many people of Aotearoa, Tongatapu. Tutuila, Ofu, Aitutaki, Rarotonga and Tahiti who helped Hōkūle'a and her crew along the way. To Dan Wright, the skipper of our escort vessel, who died suddenly a week after the arrival at Tahiti, his wife, Peg Wright, and their sons Denny and Robin, we owe special thanks for their efforts to follow in our wake, to document our passage and to stand by in case of need.