- 41

“A vanishing art” is how David Lewis characterised the traditional, Pacific Island practice of non-instrument navigation in his classic study, We, the Navigators. Actually, the systematic, non-instrument observation of stars, wind and swells in long-distance navigation has long been extinct in the main centres of the Pacific, particularly in Polynesia. Even in that last stronghold of this ancient art, the tiny and remote atolls of the Caroline Islands of Micronesia, traditional navigation is under pressure. More and more Carolinians are travelling by ship and even aeroplane instead of sailing in their own canoes, and fewer and fewer youths are willing (or able—given the demands of the modern schooling system) to undertake the many years of patient apprenticeship needed to learn how to guide a canoe from island to island without compass, chart, sextant or any other modern aids.

In recent years, however, appreciation of traditional navigation has been growing—first among students of Pacific prehistory and maritime technology, and more recently among Pacific Islanders who have lately developed an intense interest in their maritime heritage. It is in the voyages of Hōkūle'a, a reconstruction of a Polynesian voyaging canoe, that these two groups have come together to study and re-learn this vanishing art. In 1976 Mau Piailug, a master navigator from Satawal Island in the Carolines, guided Hōkūle'a from Hawaii to Tahiti, a feat which demonstrated how traditional methods can be used effectively to navigate over not just hundreds but thousands of miles of open ocean (Lewis 1977; Finney 1979). Then, in 1980, Mau's Hawaiian pupil, Nainoa Thompson, employed non-instrument methods to navigate the canoe from Hawaii to Tahiti and return. We present here an analysis of how Nainoa was able to learn how to navigate without instruments, and then apply his newly developed skills to the task of navigating Hōkūle'a over almost 5000 miles of open ocean. 1

All navigators must be able to (1) determine direction in order to set - 42 an accurate course towards their destination; (2) keep track of position en route and make any necessary course corrections; and (3) actually make landfall on the island or stretch of coast to which they are heading. To accomplish these tasks modern navigators use a variety of instruments, charts and mathematical calculations, and a spherical co-ordinate system of latitude and longitude. The traditional Pacific navigator had none of these tools, yet he was able to find his way from island to island with a remarkable degree of accuracy. How was he able to do this? The basic elements of the traditional Pacific navigation system, documented in detail by Lewis (1972), Gladwin (1970), Goodenough (1953) and others, may be sketched in terms of the navigator's three main tasks.

Stars rise in the east and set in the west, following paths which are unvarying on a human time-scale. Perceptive and practised eyes have long used these regularities for determining direction and thus for steering towards destinations far beyond the visible horizon.

To steer a course the navigator points the prow of his canoe towards the rising or setting point of the star which he has learned marks his destination. For example, if he wants to sail towards an island lying to the east, he picks the point on his mental compass marked by the star which rises above, or nearly above, the island. When sailing across wind and current the navigator picks a star course sufficiently to one side or other of the direct course in order to compensate for the estimated leeway and current drift.

Courses and directions may be denoted in terms of certain prominent stars which rise or set at the appropriate place on the horizon. These few key stars, however, will often be too high in the sky, or below the horizon, or too close to the sun, or hidden by clouds. The expert navigator must therefore know what stars rise and set at or near the same point on the horizon as the key stars so that he can use these secondary stars whenever the key stars are not visible. Furthermore, when clouds block the horizon, the navigator must be able to orient himself by glancing at any section of the night sky, much as we can by glimpsing any part of a thoroughly familiar room. For someone with this degree of knowledge of star patterns, the determination of direction is a relatively minor problem, as long as the sky is not totally obscured for extended periods.

During the day the navigator orients himself on the sun, steady winds and pattern of swells. The sun can be used for orientation when it is low on the horizon during the early morning and late afternoon, although - 43 the navigator must keep track of its changes in declination in order to be able to use it with any degree of accuracy. This can be done at dawn by noting the rising sun's position with the fading star compass. At midday, when the sun is too high in the sky to give a directional indication, or when clouds cover the sky, day or night, the navigator must determine direction and steer his canoe by reference to the winds and the more regular and enduring pattern of ocean swells.

Of all the Pacific Islanders who use stars on the horizon for orientation and steering, the navigators from the Caroline Islands have developed the most elaborate system known to us. They conceptualise a star compass in which the points are marked by the rising and setting locations of key stars (Figure 1; Gladwin 1970:149; Goodenough 1953:6; Lewis 1972:62). Because these key stars are unevenly distributed around the horizon, the Carolinian star compass looks confusingly asymmetrical; but our modern, evenly spaced orientation system is no more than a convenience which makes it easier to work with our charts, protractors, plotting tools, magnetic compasses, etc., and to do the required mathematical calculations. The continuous circle of the horizon has no inherent, naturally ordered divisions. For “eyeball” determination of direction, with no chart work and no numerical calculations, the Carolinian system—which, after all, is based on things one can see—is probably intrinsically superior, although certainly more difficult to learn.

Although a navigator may demonstrate his star compass to novices ashore by placing a circle of pebbles on a mat, the only compass the traditional Carolinian navigator takes to sea is the conceptual one engraved upon his mind through years of study and practice. 2

Although Polynesian canoe voyagers once used horizon stars for navigation, judging from the accounts of early European explorers, none of them conceptualised these stars in terms of a star compass as elaborate as that known from the Caroline Islands. Instead, from several Polynesian archipelagos there is evidence of 16-point and 32-point wind compasses wherein the direction from which specific oceanic winds blow mark the compass points.

Once at sea, a navigator has to keep track of his progress towards his destination and make any necessary course corrections. There is some evidence that Polynesians used the so-called “zenith-star” method, to be explained later, to determine their latitude in reference to particular islands, and that some Carolinian navigators judge their progress north or south in terms of the height above the horizon of Polaris and the

- 44

FIGURE 1.

The Carolinian Star Compass (after Goodenough 1953:6). The compass looks skewed to Western eyes because it is aligned on Altair (81.5°) instead of true east (90°).

Southern Cross (Lewis 1972:233-45). However, traditional Pacific navigators (and Western navigators before the time of Cook) had no way in which they could precisely fix their position on the open ocean as modern, instrument navigators can. Instead, Pacific navigators were accomplished in the art of dead reckoning, a world-wide practice whereby a navigator deduces his position from estimates of course and speed “made good” after allowing for the estimated effects of leeway (sideways drift of the vessel caused by the wind) and current.

- 45

Pacific Island navigators, however, had to develop a system which allowed them to overcome a very special handicap: without written language, their practice of navigation was entirely mental, with reckoning of position being carried forward through the voyage in a navigator's memory. The continuous level of concentration required is much higher than in modern navigation, with its written logs, and plots on charts, regularly updated.

Exactly how traditional Polynesian navigators used dead reckoning is not known. For the Caroline Islands there are abundant data on a dead reckoning system which involves mentally dividing a voyage into segments marked by the change in the bearing of an etak or “reference” island, as judged by its envisioned movement from one star compass point to another (Figure 2; Alkire 1970:52-4; Gladwin 1970:181-9;

FIGURE 2.

A short voyage of two etak segments from island A to island B (after Lewis 1972:133). When the canoe leaves island A the etak or reference island (C) “lies under” star X. When the etak island “lies under” star Y the first etak segment is completed. When the etak island “lies under” star Z the second etak segment is completed and the canoe should be at its destination, island B.

- 46

Lewis 1972:133-8). Since the etak island lies well off to one side of the course and thus cannot be seen from the canoe, the navigator conceptualises his perception of the course and speed of the canoe in terms of the changing bearing of the “moving” etak island.

This parallels the common practice of today's seamen of saying that a channel marker is “drawing abeam”, or that an island is “falling astern”, when it is obvious that it is the vessel, not the channel marker or island, that is moving. The modern sailor only applies this personal viewpoint to objects he is looking at, however. When he considers objects over the horizon, or when he thinks about the voyage in abstract, he normally switches to a “God's eye” view, looking down on his chart as though looking down on the ocean from a great height, visualising the fixed islands and the moving vessel's position.

The Carolinian navigator, not familiar with the concept of representing the earth's surface on a chart, retains the personal viewpoint even for objects he cannot see. Looking outward from the canoe with his mind's eye, he visualises the etak island far off to one side, changing its bearing as the canoe sails towards its destination. The modern navigator reads his instruments which show the progress of his vessel, and then mathematically converts these into miles in a particular direction to add to the course plotted upon a chart. When the proper number of miles have been run off in the right direction, the destination should appear. His traditional Carolinian counterpart intuitively converts his visual observations of the canoe's progress into estimates of changes in the bearing of the etak island. When the bearing marking the completion of the final segment of the voyage is attained, his canoe should be off the target island. Both systems use mental constructs which cannot actually be seen; both systems work.

Small islands, particularly low atolls, are easily missed at sea if all one relies upon is the visual sighting of the island itself. The tops of an island's tallest coconut palms provide a navigator with the first visual sign of an atoll, but not until his canoe is within 10 miles or so. In order to expand the detection range of such easily missed targets, Pacific navigators look for signs of islands ahead in the clouds, the patterns of the ocean swells and the flight of birds. High island peaks may interrupt the flow of trade wind clouds, causing a noticeable piling-up effect long before the peaks themselves can be seen. Shallow lagoons may cast a reflection upon the undersides of passing clouds. Ocean swells bouncing back from an island obstacle ahead, or curving around it, provide subtle - 47 clues for another remote sensing technique, but one which takes great experience to apply. In most areas the favorite landfinding aids are those birds—primarily the noddies, terns and boobies—which sleep on land but fly out to sea each day to fish. Navigators searching for land welcome the sight of these birds, for typically they fly out only as far as 20 or 30 miles, and if spotted returning home at dusk can even provide the navigator with an accurate bearing to the island.

The traditional navigator's task of finding land is further lightened by the patterning of the oceanic islands, caused by the interaction of the Pacific lithospheric plate and “hot spot” sources of magma beneath it. As this plate has slowly moved to the north-west, lava from these hot spots has periodically punched through the ocean floor to form long series of oceanic islands. Conveniently for the navigator who happens along much later, these chains of volcanic islands or their atoll remnants furnish much larger and easier to find targets than any lone island. A navigator heading for one island in the chain can intercept the chain at any point in its length, reorient himself, and then proceed to the particular island target.

Hōkūle'a, a 19-metre-long double-canoe, was built to test the sailing ability of a voyaging canoe, and the feasibility of navigating over long distances by traditional methods (Finney 1979). The design of the canoe was based on a comparative analysis of traditional Polynesian canoe types as recorded by early European navigators and artists, and was an attempt to recreate the lines of an Eastern Polynesian voyaging canoe of around the 12th century. The intent was to replicate the probable performance of such a canoe; for practical reasons, however, construction materials and methods were mostly modern.

The round-trip route between Hawaii and Tahiti was chosen because Hawaiian legends told of voyaging back and forth between these Polynesian centres, and because there were some archaeological and linguistic indications of Tahitian-Hawaiian contact (Finney 1967). However, as non-instrument navigation had long since disappeared from Polynesian waters, the Carolinian navigator, Mau Piailug, was recruited to guide Hōkūle'a. Even though some of his methods might differ in detail from Polynesian methods, it was felt that Polynesian and Carolinian navigation were sufficiently similar (and indeed historically related) to make his participation in a primarily Polynesian experiment valid (Lewis 1977).

Mau was not familiar with the islands, winds and currents along the - 48 route, however, nor had he ever sailed in latitudes as far north as Hawaii or as far south as Tahiti. Inasmuch as the object was to retrace a route over which, according to tradition, voyagers had travelled back and forth many centuries ago, and not to recreate a voyage of discovery, Mau was briefed on the geography, sailing conditions and changing skies along the route—all things which a traditional Polynesian navigator sailing over that course would have known. This was accomplished through discussions in Hawai'i, sessions at the Bishop Museum Planetarium there, and briefings en route by individuals who had already sailed in those waters.

The route between Hawaii (main islands 19° to 22°N) and Tahiti (18°S) crosses three main wind and current zones. In 1976 Hōkūle'a sailed to Tahiti during the months of May and June; in 1980 the voyage was made during March and April. The following descriptions are of typical conditions along this track during the months of March to June. 3

• 1. From Hawaii to approximately 9°N—the zone of the north-east (NE) trade winds and westerly setting North Equatorial Current. The average wind direction is actually about east-north-east.
• 2. Approximately 9° to 4 °N—the Intertropical Convergence Zone (ITCZ), with the eastward setting Equatorial Countercurrent. This region of convergence between the NE and south-east (SE) trades is characterised by a high frequency of light and variable winds, frequent intervals of heavy cloud cover, and high rainfall, sometimes in sharp squalls. Belts of calm (“doldrums”) develop seasonally along the ITCZ in the central Pacific, with greatest prevalence in the northern summer. The average latitude of the centre of the ITCZ is shifting northward from 4°N in March to 7 °N in June, and the prevalence of calms is increasing, but the whole zone is subject to rapid short-term variability in winds and cloudiness. The speed with which this region can be transited under sail depends heavily on luck.
• 3. From about 4 °N to Tahiti—the zone of the SE trade wind and the westerly setting South Equatorial Current. The average wind direction is east, increasing gradually in strength during the months from March to June.

Since Hawaii lies several hundreds of miles to the west of the meridian of Tahiti, well to leeward in reference to the trades, the main sailing problem is to make enough distance eastward while sailing across the NE trades, and then, sailing against the SE trades, to hold far enough to the east so as to reach Tahiti and not be pushed to the west of that island by wind and current. In 1976 Hōkūle'a proved capable of doing this. Landfall was made on the 32nd day at sea on Mataiva Atoll in the Tuamotus; - 49 from there the canoe was easily sailed to Tahiti (Figure 3; Finney 1977, 1979).

FIGURE 3.

The 1976 voyage of Hōkūle'a to Tahiti showing main wind and current zones (after Finney 1979:223).

- 50

For most of the voyage to Tahiti the canoe was sailed as close to the wind as was possible without losing too much speed. This need to push hard into the trades meant that the navigational task was not primarily one of picking the appropriate star course and keeping the canoe on it, but rather the more passive process of monitoring, by reference to the stars, wind and swells, where the wind allowed the canoe to sail. Only when there were light variable airs in the ITCZ was there any real freedom in choosing the course. Because of the heavy overcast and the confused seas encountered there, however, this proved to be a most difficult area in which to navigate, although Mau managed to find his way.

Sailing in strange seas over a totally unfamiliar route many times longer than any voyage he had undertaken before presented Mau with a formidable problem of adapting his etak dead reckoning system. Based on the briefings he had received, and his study of the charts, Mau chose two archipelagos to serve as etak islands: the Marshall Islands far to the west in Micronesia for the portion of the voyage north of the equator; and for the portion south of the equator, the Marquesas Islands located a few hundred miles to the east of the projected route. Unfortunately, how Mau handled the relatively large etak segments that resulted from this arrangement was not precisely recorded, nor is it clear how much Mau may have used the changing elevation of the stars on this long north- south route to help him in his dead reckoning. None the less, it was documented that just before landfall in the Tuamotus Mau accurately estimated the canoe's position in terms of the bearing to the Tuamotus and Tahiti and the approximate sailing time to them.

Had Mau navigated Hōkūle'a back to Hawaii, much more would undoubtedly have been learned about how he applies the etak system and his other navigational methods. Unfortunately, crew problems led him to fly directly home to Micronesia, and the canoe was navigated back to Hawaii by modern methods. This loss, plus several unfortunate incidents during the first leg of the voyage that threatened the integrity of the navigational experiment, limited the impact of this effort to test traditional navigational methods. Hence, a replication of the experiment was called for: a repeat voyage on which Nainoa would navigate the canoe to Tahiti and then, to complete the original experimental plan, would navigate her back to Hawaii.

Nainoa, then a university student, sailed back to Hawaii in Hōkūle'a. When he had first been recruited as a prospective crew member in 1974, - 51 Nainoa had no knowledge of Polaris or other key Northern Hemisphere stars, and no idea that the Southern Cross could be seen from the south shore of his home island of O'ahu—despite the fact he had spent years surfing and fishing there. As Nainoa learned more about the plans to sail Hōkūle'a to Tahiti by non-instrument methods, and was shown how to spot Polaris, the Southern Cross and other stars and constellations useful for navigation, he became fascinated with the idea of traditional navigation. He took an astronomy course at his university, and began reading astronomy books and the works of Lewis, Gladwin and others who had written about Pacific navigation. Although Nainoa did not sail on the Hawaii to Tahiti leg, and therefore had no opportunity to see Mau in action, his enthusiasm for non-instrument navigation was further reinforced by the experience of sailing Hōkūle'a back to Hawaii. During that voyage he had an opportunity to practise some of the techniques he had read about, such as steering on horizon stars and judging latitude from the height of Polaris. He also began to develop some unique ways of his own for reading the stars.

After returning to Hawaii in August 1976, Nainoa continued to study the sky and the sea, taking every opportunity he could to observe the

Nainoa Thompson aboard Hōkūle'a. (Will Kyselka)

- 52

stars, winds and swells when sailing in Hōkūle'a and when going out to sea in his own fishing launch. His approach to the night sky went far beyond the necessary memorisation of the stars, their rising and setting points and their declinations. Nainoa was always searching to discover patterns and relationships which would enable him to determine direction precisely and help him keep track of his position at sea.

When, in early 1977, Nainoa was puzzled by a problem concerning the rising of the sun and the moon, he asked for help at Honolulu's Bishop Museum Planetarium. From that time on, Nainoa and planetarium lecturer Will Kyselka spent literally hundreds of hours in the planetarium studying the sky at various latitudes, seasons and time of the night, and investigating the hypotheses Nainoa formulated on how the stars, moon and even the planets might be used to navigate Hōkūle'a on a second voyage to Tahiti.

A premature attempt to sail Hōkūle'a to Tahiti in 1978 was cut short when the canoe capsized in stormy seas soon after leaving Honolulu. While the canoe was being repaired, and plans for another attempt to return to Tahiti were being formulated, Nainoa continued his studies, going so far as to fly to Tahiti in order to acquaint himself with how the stars looked at that latitude. Then, still feeling that he did not know enough about traditional navigation, Nainoa flew to Saipan in the Northern Marianas group of Micronesia to see Mau, who had just sailed his canoe there from Satawal, 500 miles to the south. After hearing Nainoa's earnest appeal for help, Mau accepted the young Hawaiian as his pupil, and in September of 1979 flew to Hawaii to begin tutoring Nainoa.

The task was formidable, for time was short. Mau's own apprenticeship had lasted a dozen years. Only after innumerable sessions on the beach at Satawal, and voyage after voyage made under the watchful eyes of senior navigators, was Mau formally initiated as a navigator and thus in effect licensed to practise his craft. Although Nainoa was a highly motivated pupil with three years of intensive self-study behind him, the problem was that departure for Tahiti was scheduled for March 1980, just six months away.

The first month was spent on land in formal sessions held at Nainoa's house. Mau would take 32 pebbles, arrange them in a circle to represent the star compass, then take long, slim leaves stripped from a coconut frond to represent bearing lines from a model canoe to star points, and to simulate the pattern of swells passing under the canoe. Nainoa was expected to memorise the compass points, the stars for which they are named, and the name and direction of the dominant swells. After each - 53 session Nainoa and Mau would drive around the island of Oahu seeking vantage points for each sector of the sky so that Mau could point out key navigational stars.

After a month of these intensive sessions, by which time Nainoa was able to pick out the compass points and identify the stars with ease, the two started going out to sea in Nainoa's boat. There Mau continued the lessons on how to read the sky and swells for direction, and also showed Nainoa which birds were the best indicators of land.

Mau was an exacting teacher. He expected Nainoa to take the information given him, digest it, think about it, and then come back with further questions—but not about the identification of particular stars or details which Mau felt he had already taught his pupil. As a man from a non-literate culture where memory skills are highly developed, Mau expected Nainoa to grasp immediately the information given him, and then to come up with questions which showed that he was thinking ahead and anticipating how he might apply what he had learned. In addition to these structured sessions, Nainoa also learned a great deal from Mau when the two relaxed together, drinking beer and talking for hours about sailing and navigation.

This pedagogical method, both in its formal and informal aspects, worked well for many facets of Mau's navigational system, but not for imparting a working knowledge of etak dead reckoning. Mau briefly explained to Nainoa how star bearings over an etak island are used to keep track of a canoe's progress along a course line, but Nainoa never felt that he understood etak reckoning well enough to use it. Instead, he concentrated on developing his own dead reckoning system, to be described in the following section. It was based on Mau's teaching but was essentially Nainoa's construct.

By the early spring of 1980 Nainoa felt that he was ready to try to navigate Hōkūle'a to Tahiti; Mau also believed that Nainoa was prepared for the task.

The system Nainoa developed is not specifically Carolinian, or Polynesian, or modern. It is a unique mixture made up of traditional Pacific methods blended with techniques Nainoa has worked out for himself. These techniques were not adapted from modern celestial navigation, which Nainoa deliberately avoided learning, but grew out of his own study of the stars in astronomy books, in the night sky, and at the Bishop Museum Planetarium. None the less, however idiosyncratic - 54 Nainoa's navigation system may be, he built it on a traditional Pacific base and employs it at sea without the use of any instruments, charts or written materials.

Like traditional navigators, Nainoa uses horizon stars, plus Polaris, the Southern Cross and other circum-polar stars to frame a star compass. However, although his compass has 32 points like the Carolinian one, it differs from its traditional model. Where the points of the Carolinian compass mark the exact rising and setting points of key stars, and are thus unevenly spaced along the horizon, in Nainoa's compass the points are evenly spaced along the horizon regardless of the rising and setting points of the navigational stars (Figure 4).

FIGURE 4.

Nainoa Thompson's Star Compass showing the Hawaiian names he uses and their English equivalents.

- 55

Where the Carolinian compass is aligned on the star Altair, which rises 8.5° north of east, Nainoa uses true north and south, marked by Polaris and the long axis of the upright Southern Cross, with east and west at right angles to this. Nainoa calls these cardinal directions by their traditional Hawaiian names: 'ākau (north), hema (south), hikina (east), and komohana (west).

Nainoa further refined his compass by halving each quadrant formed by the cardinal points, then by halving the resultant sectors twice more to finally come up with a 32-point compass with equal sectors of 11.25° each. Nainoa calls each sector a “house”, and for his own reference calls them by Hawaiian names he has arbitrarily assigned to them: manu (bird), lā (sun), 'āina (land), noio (noddy tern), haka (empty), nā leo (the voices), nā lani (the heavens). Nainoa carries this compass in his mind, along with the locations of the many stars he uses for navigation. Thus, whenever he can see the stars Nainoa can mentally impose his compass around the horizon.

Many readers may note that Nainoa's compass matches the original 32-point Western mariner's system, used for compass directions and for bearings relative to a vessel (“two points on the port bow”, etc.). Nainoa was unaware of this, as he was familiar only with the modern system in which compass directions are given in degrees clockwise from north. He expressed surprise when he discovered that he had, in effect, re-invented a form of the wheel! (Figure 4).

On cloudy nights when no stars can be seen Nainoa orients himself the winds and ocean swells, a practice which at times can be most difficult. In the morning and late afternoon he uses the sun for orientation; and watches the wind and swells at midday.

For those hazy, partially overcast nights when the stars are difficult to make out, but the moon or planets can be seen, Nainoa uses methods he himself has developed to obtain rough bearings. His “cut of the moon” is an imaginary line through the horns of the crescent or quarter moon; it yields a north-south line when the sun is on the celestial equator, but progressively deviates from that line as the sun moves north or south during the seasons. By keeping tabs on the changing angle of the cut of the moon in relation to his star compass during periods of clear skies, Nainoa is able to get a rough approximation of direction on partially overcast nights. Similarly, Nainoa keeps an eye out for the changing positions of the bright planets and their relation to the star field so that they also may serve as directional guides on hazy nights.

- 56

Although Nainoa was not able to learn Mau's etak techniques, he was able to develop his own dead reckoning system based on Mau's instruction that, for the Hawaii to Tahiti route, Nainoa should construct a reference course line that takes into account the expected wind and current conditions, and then, during the voyage, should continually keep track of the canoe's changing position in relation to that line.

Nainoa implemented this instruction by developing a way of picturing his changing position at sea in terms of compass “houses”. When, for example, Nainoa judges that the wind has forced the canoe off the course line, he estimates how many compass points, or “houses”, the canoe has deviated from the course and how far it has sailed in that direction. For example, he would say that he is three houses west of a reference course of due south. This means that he reckons that the canoe is one day's sail at three compass points west of the course on a heading of south-west by south. The utility of this method is that it enables Nainoa to summarise a mass of observations and estimations of the canoe's speed and heading and express the result in one number. Thus, instead of having to remember that the canoe sailed, for example, 25 miles in one heading, then 50 miles in another heading, and so on throughout the entire voyage, Nainoa can periodically reckon his position in terms of how many houses the canoe is off the reference course, and carry this single number over to the next day's reckoning.

Nainoa has also tried to develop the practice, inspired by Mau's instruction, of keeping track of the changing bearings of the departure island, the target island and any islands to one side or another of the course, although he does not feel confident of his ability to make very accurate estimates.

In addition to these ways of estimating position and bearing in relation to his reference course and to key islands, Nainoa also employs a number of methods for making fairly precise estimates of his latitude.

The angular height of Polaris above the horizon is virtually equal to the latitude of the observer, a relationship that early navigators from Europe and Asia used to their advantage. Thus, if Polaris appears to be 10° above the horizon, the observer's latitude is approximately 10° north of the equator. Nainoa developed a way of measuring the height of Polaris by holding his right arm out, placing his outstretched thumb along the horizon and then gauging where Polaris intersects his upright fingers. For example, O'ahu lies just over 21° north of the equator. - 57 Polaris, which from O'ahu appears elevated a little over 21° degrees above the horizon, hits Nainoa's hand just at the top of his index finger. By experimenting in the planetarium and the night sky, Nainoa has been able to calibrate his hand so that he can make a fairly accurate estimate of the height of any star which lies fairly close to the horizon. (Lewis, 1972:242, reports that some Carolinians used a similar method, although Mau does not.)

Nainoa also determines latitude by using pairs of stars, each pair chosen so that the two stars lie approximately north and south of each other on the celestial sphere, and are fairly close together. Many such pairs are available because this system of navigation is not limited to the relatively few stars bright enough to be visible at twilight for sextant observation.

In his study Nainoa learned that when such a pair appears vertically upright in the sky it marks the observer's celestial meridian, and thus true north or south. He then realised that he could obtain latitude from such pairs as accurately as he could from Polaris, requiring only that he memorise the angular height above the horizon which the bottom star of each pair would have, with the pair upright, if observed on the equator. (In astronomical terms, this quantity is the “polar distance” of a star.) When an upright star-pair is observed, the latitude of the observation is given by the difference between the (measured) height above the horizon and the (memorised) height it would have if seen from the equator. Nainoa also memorised one further number relating to each star-pair: the angular separation between the stars of the pair.

The use of star-pairs has two advantages. First, the position of the celestial meridian is clearly marked by a pair's upright position; no other direction or time clues are needed. Second, the measurement of angular height above the horizon is made easier by the ability to compare the measurement with the (memorised) angle between the stars of the pair. In effect, the pair acts as a calibration for the measurement of height.

As an example, the separation between the top and bottom stars of the Southern Cross is 6°, while the bottom star of the Cross (when upright in the sky) would be approximately 27° above the horizon if seen from the equator. If the height of the bottom star of the upright cross were observed to be 17°, for instance, the latitude of the observation would be 10°N (27°-17°=10°).

When the relationship between the pair-angle and the height of the pair above the horizon is simple, Nainoa can use this “proportional spacing” to determine latitude almost at a glance. For an important example, at the latitude of Honolulu the bottom star of the upright - 58 Southern Cross is just midway between the horizon and the topmost star of the Cross (6° from the horizon to bottom star; 6° between bottom and top star). It is much easier to see such an equality of spacing than to measure the angular height of any single star, such as Polaris. The Southern Cross was, in fact, one of the most valuable latitude indicators as Hawaii was approached during the return voyage.

Still another, though more restricted, method Nainoa developed for measuring latitude involves the simultaneous rising or setting of two stars. In his nocturnal observations he discovered that two stars will rise simultaneously at only one specific latitude. For example, at 21° North Latitude Arcturus (Hōkūle'a in Hawaiian) and Spica rise together. But, as one travels southward Spica rises progressively earlier than Arcturus. Hence, sailing north from Tahiti Nainoa could roughly monitor his progress homeward by watching the gap between the rising times of the two stars grow smaller and smaller.

Nainoa has also tried the zenith star method for determining latitude, but with less success than David Lewis has obtained in his experiments. Lewis (1966, 1977) found that he could obtain a fairly accurate measure of latitude by sighting up the mast to identify stars which passed through his zenith (the point directly above the observer; an observer's latitude is equal to the declination of the star at its zenith). The accurate determination of the zenith point by eye alone is difficult even on land. Nainoa found it doubly so at sea, and discarded this method in favour of using Polaris and meridian star-pairs.

Of the three main methods for expanding the sighting range of a single target island, Nainoa has focused primarily on learning how to spot the appropriate landfinding birds. He does not rely on the observation of cloud patterns, and he finds that he has yet to develop the skill needed to discern those disruptions in the swell pattern which signal proximity to land. His overall landfinding strategy of course makes full use of the concept of aiming for a wide group of islands, rather than a single point.

The 1980 voyage of Hōkūle'a to Tahiti and back was designed to replicate and extend the navigational experiment carried out on the 1976 voyage. Nainoa proposed to replicate Mau's feat of navigating to Tahiti, and then to extend it by navigating back to Hawaii, all without in- - 59 struments, charts or other modern devices. This navigational experiment was to be more than just a replication and extension of the previous one, however. For all his hard work, Nainoa was well aware that he had very little experience in comparison with someone like Mau. The 1980 voyage was therefore also an extraordinarily demanding test for a novice navigator attempting to recreate the legendary navigational feats of his ancestors.

Mau sailed on both legs of the voyage, with the understanding that he would not intervene in the conduct of the voyage unless the safety of the vessel was at stake, or unless Nainoa had become lost or requested his aid.

Mau's skill as a deep-water canoe seaman proved invaluable in many unexpected ways. For example, during stormy weather at the very beginning of the voyage Mau did intervene in the sailing of the canoe by ordering the lowering of the sails upon the approach of heavy squalls. In addition, he repaired the boom on the forward sail when it broke. However, he did not intervene in the navigation of the vessel, and only once—just before Hawaii was sighted on the return voyage—did he query Nainoa about a navigational decision. Nainoa did occasionally seek Mau's advice on such difficult problems as how to read direction from a complex field of swells, and how to predict impending wind shifts from cloud formations. In this way the voyage served as a continuation of Nainoa's navigational education, as well as a test of what he had learned so far.

Gordon Piianaia captained Hōkūle'a and Leon Sterling served as mate. In addition, eight men and two women sailed as crew on the Hawaii to Tahiti leg, and 10 men served as crew on the return to Hawaii. 4 One of the crew members who sailed on both legs, Stephen Somsen, documented Nainoa's navigational effort, primarily through a regular schedule of tape-recording Nainoa's position estimates, his navigational decisions, and his comments on these matters and on sailing conditions. Nainoa's position estimates were composed of two parts: (1) latitude estimates; (2) estimates of distance and direction from the reference course line phrased in terms of “houses”. Only later in Hawaii were these estimates plotted on a chart so that they could be compared with the actual position of the canoe at the time Nainoa made his estimates.

An accurate track of Hōkūle'a was constructed through the use of the ARGOS system of position determination, a co-operative project operated by the Centre National d'Études Spatiales (the French national space agency) and the National Aeronautics and Space Administration and the National Oceanic and Atmospheric Administration of the United States.

- 60

This system is widely used by oceanographers to track drifting unmanned buoys, by meteorologists to track balloons, and by biologists to track such animals as bears and sea turtles. Because the object being tracked carries only a small, automatic radio-beacon the ARGOS system was ideal for the navigation experiment.

The “Platform Transmitter Terminal” (PTT), as the radio-beacon is called, which the canoe carried was lent by Dr William Patzert (then of the Scripps Institution of Oceanography, University of California; now with NASA), who was involved in oceanographic studies of the central equatorial Pacific at the time of the voyage. The PTT was turned on and sealed, together with its battery pack, in a waterproof housing before departure for Tahiti. A new battery pack was installed during the canoe's layover at Tahiti; otherwise the PTT was lashed down securely for the duration of the voyage, requiring no attention by the crew.

During operation the PTT continuously transmits a series of precise signals which can be received by either of a pair of satellites in polar orbit when that satellite is within range. The satellite records signals from the PTT which allow later identification of the individual unit and calculation of its position. When the satellite subsequently passes over one of the special ARGOS ground-stations, the recorded information is sent down and relayed to the ARGOS system centre in Toulouse, France. The Toulouse centre computes the PTT position at the time of each satellite pass, and then makes this information available to users within a few hours of the satellite's acquisition of the PTT signal.

An average of six position fixes were obtained for Hōkūle'a each 24 hours during the voyage, with position accuracy within a few hundred metres, and without any possibility of information being returned to the canoe to compromise the navigation experiment.

The yacht Ishka, owned and captained by Alex Jacubenko, escorted Hōkūle'a to Tahiti and return both for safety and documentation. During daylight the Ishka sailed a mile or two behind the canoe, sometimes to leeward, sometimes to windward; at night it closed in so as not to lose the canoe in the dark. To ensure that records of Nainoa's position estimates and other navigational information would not be lost, data were transmitted to the Ishka by means of a hand-held VHF transceiver that allowed communication within line-of-sight. Nainoa did not participate in these transmissions; nor was any navigational information sent from the escort vessel to the canoe except after land was sighted.

Before leaving Hawaii Nainoa constructed a course line to serve as - 61 a reference against which he could estimate the changing position of Hōkūle'a throughout the voyage. First he drew a heading line, based on an estimate of average wind direction and how close the canoe could sail against the wind, through the three wind and current zones. Then he altered the resultant dog-leg line according to how much he estimated the current would set the canoe to one side or another. For example, to calculate the reference course for the NE trade wind zone, he started with a heading of south-east, multiplied the estimated daily current set to the west of 15 miles times an estimated nine days to be spent sailing through that zone to come up with a total current set of 135 miles to the west. Then he moved the end point of the heading line 135 miles to the west and redrew the line to make a reference course which was somewhat to the south of south-east. Figure 5 shows the reference course to the Tuamotus, the actual track of the canoe according to satellite fixes, and the location of Nainoa's dead reckoning (D.R.) fixes. Figures 6 and 9-11 indicate for successive segments of the voyage the actual track of the canoe, Nainoa's D.R. positions and the exact relationship between the two. A summary narrative of the voyage, divided according to the three wind and current zones, follows below.

Before beginning that narrative, however, a word about the sails that drove the canoe to Tahiti and return is in order. Hōkūle'a is rigged with “crab-claw” sails patterned on petroglyphs of traditional Hawaiian sails, and on drawings of 18th century canoes made by early Western visitors. In this rig the sails are permanently bent to a yard and boom. The yard stands vertically against the after side of the mast, while the deeply curved boom sweeps up to approach the vertical at its extreme end, reaching nearly as high as the yard. The free edge of the sail is along the top.

The canoe has two sails, a larger, forward sail, and a smaller, after sail. Going to windward, the canoe can be made to steer herself by trimming the sails properly. Off the wind it is necessary to steer with a long steering paddle or sweep.

To shorten sail temporarily, the boom is “triced” up to the yard, with the two spars together like the pincers of a crab's claw. This leaves a lot of weight and windage aloft, so that for longer periods, or during heavy weather, the entire rig (yard, sail and boom) is lowered to the deck, leaving only the masts standing. The spars are heavy, however, and no winches or blocks are used, so lowering the rig can be both difficult and dangerous.

- 62

FIGURE 5.

The 1980 voyage of Hōkūle'a from Hawaii to Tahiti showing actual track of the canoe, reference course, and dead reckoning (D.R.) positions made (starting March 20) by Nainoa Thompson. The number beside each D.R. position refers to the date: “a” for estimates made before noon (often at sunrise); “b” for estimates made after noon (often at sunset). Approximate current boundaries from aircraft expendable bathythermograph observations made along 150°W, April 8, 1980 (Stroup et al., 1981).

- 63

On March 15 Hōkūle'a left the port of Hilo on the east coast of the island of Hawai'i. 5 The yacht Ishka towed the canoe for several hours until the offshore trades were reached. The sails were then let out and Hōkūle'a started sailing on the port tack (wind coming from the left side) as close towards south-east as possible given the easterly angle of the NE trade winds.

Although the weather had looked favourable upon departure, the already strong trades soon began to intensify and the sky clouded over. By that evening the wind had increased to at least 30 knots, and as a precaution against squalls the sails were “triced”. By the next morning the wind was blowing at 40 knots with stronger gusts; just before noon the sails and spars were lowered to the deck and the canoe drifted under bare poles until the following morning, by which time the winds had moderated. The sails were then raised and Hōkūle'a set off on the star-board tack (wind coming from the right side), sailing slightly east of north in order to make up for some of the miles lost drifting to the west. By the next day, March 18, Nainoa reckoned that they were near the reference course line, so he had the canoe put back on the port tack to resume the drive to the south-east.

Hōkūle'a stayed on the port tack, driving into the trades, for the next week. This period saw clearing skies and moderating trade winds, and went without incident except for a problem with the long curved boom of the foreward sail. During the night of the 19th, while lowering the sail just before a squall, the boom was cracked when it was accidentally let down upon the hut built on the canoe's central platform. After Mau lashed the cracked area, the sail was raised; it held until the 23rd when, while sailing in moderate winds and clear skies, it broke clean through in the same place it had previously been cracked. While Hōkūle'a drifted, Mau took a spare piece of wood, deftly shaped it to replace the broken area, carefully scarfed the joints, and then lashed the new piece securely into place. (Mau's freehand adze work was a revelation to the crew, most of whom had grown up using power tools.) The sails were then raised and the canoe started moving once again towards the south-east.

The stormy weather encountered upon leaving Hawaii, the enforced drift, and then all the trouble with the boom made an already difficult start to the navigational task all the more trying. In a tape recording of his impression of those first days, Nainoa complained, only half in jest: “Too many clouds. No stars. The wind is too strong from the east. We're always pulling the sails down drifting all over the Pacific Ocean. It's raining all the time”.

- 64

FIGURE 6.

The 1980 voyage of Hōkūle'a (Hawaii to 9° N) showing actual track and Nainoa Thompson's D.R. positions. The number beside each D.R. position refers to the date: “a” for estimates made before noon; “b” for estimates made after noon. The dashed line indicates the relationship between the D.R. position and the true position of the canoe at the time.

- 65

During this initial period Nainoa had a general idea of where the canoe was in terms of the bearing and distance back to the island of Hawai'i, but he was not able to make the star sightings he needed in order to make a more precise estimate of position. Not until the night of March 19-20 did the sky clear sufficiently to enable him to get a good look at the stars. That night Nainoa estimated that Hōkūle'a was almost 15° north of the equator, and that after all the drifting and tacking the canoe had ended up three “houses” west of the reference course line. Comparison of this position estimate and the actual track indicates that Nainoa had placed the canoe about 106 miles to the south-east of where it actually was at the time (Figure 6).

After a day or so of sailing under clearing skies and steadying trade winds, Nainoa's position estimates came much closer to the mark. In fact, on the morning of March 22 he placed the canoe within about five miles of where it actually was. Although Nainoa was not able to keep up this degree of precision, a comparison of all of Nainoa's D.R. positions with the actual track of the canoe shows that his dead reckoning was quite accurate during this fair weather period.

The clearing skies after the storm enabled Nainoa to make multiple star observations in order to determine latitude. For example, on the night of March 23-24 he was able to estimate that the canoe should reach 10°N around sunrise by measuring, with his outstretched hand, the angular height of Polaris (Figure 7) and the angular height of Acrux (the

FIGURE 7.

Nainoa Thompson's method of using his hand to measure the angular height of Polaris in order to determine latitude (night of March 23-24, 1980).

- 66

bottom star of the Southern Cross when upright, Figure 8). In addition, observations of two other stars confirmed the latitude.

This measurement turned out to be only 21 miles off, for at sunrise the canoe was actually 21 minutes of latitude north of 10 °N. At this time, however, Nainoa's reckoning of the canoe's course began to diverge from the actual course; after several days during which his D.R. positions had been close to the actual track of the canoe, he now began to overestimate the amount of eastward distance the canoe was making, in effect placing the canoe about 60 miles to the east of where it actually was.

During the transit of the NE trade winds Hōkūle'a was being sailed as close to the wind as practicable. In other words, the canoe was being sailed on a wind course rather than a star course. As was the case in 1976, when the canoe was sailed on a similar wind course, for much of the time the steering paddle was pulled out of the water and the vessel was trimmed—by shifting stores and adjusting sails—to sail automatically to

FIGURE 8.

Nainoa Thompson's method of using his hand to measure the angular height of Acrux to determine latitude on the night of March 23-24, 1985. With the Southern Cross in this upright position Acrux is 10° lower than it would be if seen from the equator, which means that the canoe must be 10° north of the equator. Nainoa was also aware that at Honolulu (21 °N) Acrux is 6° above the horizon when the Cross is upright, a height just equal to the 6° vertical span of the Cross itself, and thus especially easy to see. On the evening illustrated above, Acrux is 11 ° higher than at Honolulu, meaning that the canoe must be 11° of latitude closer to the equator than Honolulu, i.e., at 10°N.

- 67

windward. Nainoa's job was then not one of setting a star course for a steersman to follow, but of keeping track of the heading of the canoe as it drove into the NE trades.

This more passive approach to navigation did not mean that Nainoa could sit back and relax. To the contrary, he had to be constantly alert to detect changes in heading resulting from wind shifts. To sleep even for a couple of hours at a stretch could mean missing a critical wind shift, or a needed star sight. Accordingly, Nainoa tried to limit his sleeping to 20-minute naps taken on the bare boards of the canoe's central platform. To guard against oversleeping, Nainoa would rest his head on the raised central spine of the deck; after 20 minutes or so of balancing on that hard pillow his head would slip off, waking him. In this way, although he might sleep for two or three hours during any 24-hour period, Nainoa could be sure that he was only briefly absent from the watch over the wind, swells and stars. This near-constant vigil is required of every traditional navigator because he must keep a running picture of the canoe's progress in his mind. To sleep for long periods, and thus miss observing big chunks of a voyage, could critically degrade the accuracy of that picture.

Nainoa would have liked to have seen the canoe's twin prows framing Antares, the prominent reddish star in the constellation Scorpio, which rises in the early evening at this time of the year. That might have been possible, had the trades come straight out of the north-east. Since Antares breaks the horizon at a point about 26° south of east, this would have meant that the canoe was, after making allowance for leeway and current drift, on the desired course to the south-east, or even a bit to the east of south-east. Just when a clear eastern horizon was needed, however, Nainoa found that, night after night, low trade wind clouds blocked his view. This forced him to judge the canoe's heading, a little less precisely, from the bearings of Polaris and the Southern Cross—which showed that the trades, coming out of the east-north-east as they usually do, had put the canoe on to a heading somewhat to the south of straight south-east.

At dawn on March 25, almost 10 days out of Hilo, a cloud bank loomed to the south-west. Within a few hours it began to rain and the wind died. Hōkūle'a had entered the ITCZ (Figure 9).

For the next four days the canoe experienced alternate periods of calms, punctuated by short but sharp squalls, and spells of variable

- 68

FIGURE 9.

The 1980 voyage of Hōkūle'a (9 °N to 1 °S) showing actual track and Nainoa Thompson's D.R. positions. The number beside each D.R. position refers to the date: “a” for estimates made before noon; “b” for estimates made after noon. The dashed line joins the D.R. position with the true position at the time. Approximate current boundaries from aircraft expendable bathythermograph observations made along 150°W, April 8, 1980 (Stroup et al., 1981).

winds. At first, whenever the wind was strong and steady enough, the canoe was sailed to the south-east. Then, on the 28th, a brisk south wind sprang up which allowed the canoe to be sailed due east on the starboard tack. By sunrise on March 29 the wind had shifted back to the east, and the canoe was put back on the port tack to hurry southward out of the ITCZ. By that evening the stars came out clearly, marking the exit from the ITCZ; Hōkūle'a had spent only four and a half days in this zone - 69 compared with the full week the canoe had been forced to spend there during the 1976 voyage to Tahiti.

For much of this period in the ITCZ high overcast and lower layers of clouds covered the sky, obscuring the stars at night and even making it difficult to make out the bearing of the sun during the day. In the absence of star and sun sights, and steady trades as well, Nainoa could only look to the swells for orientation. Fortunately, after the confused seas encountered during those first stormy days out of Hilo, the swell pattern had become progressively more regular as the canoe sailed southward. Particularly prominent was a strong swell from the northeast which persisted into the ITCZ, enabling Nainoa to orient himself there in the absence of other clues.

However, the lack of star sights for latitude determination, and the irregular winds, made it tremendously difficult for him to estimate the canoe's progress through the ITCZ. A comparison of his D.R. positions with the actual track indicates that, although he was able to discern the general trend of the canoe's course between March 25 and March 29, the sequence of his position fixes in between would seem to indicate that he thought the canoe followed a much more erratic course than it actually did, and that on two occasions it even reversed direction. This appearance is misleading, for the seemingly erratic pattern of the D.R. positions only reflects how Nainoa changed his mind about the canoe's progress during this navigationally difficult period. In no way do these points indicate that Nainoa thought the canoe actually sailed from one estimated position to another. (D.R. positions in all the figures have been left as unconnected points to avoid the implication that Nainoa actually thought the canoe moved from one position to another.)

Once out of the ITCZ and under clear skies, Nainoa was able to recalculate his position with some accuracy. For example, on the night of March 29-30 his first star sights made him revise his previous latitude estimate of 3.5 °N (position 29b, Figure 9) to 3 °N (position 30a, Figure 9), a position much closer to the actual.

The light easterly winds encountered upon leaving the ITCZ continued until April 2 when, at about 2 °S, the wind shifted to east-southeast. The wind held this general direction, and picked up strength, over the next several days. Then on the 6th the wind began to drop, and for the next five days it was extremely light. “Nainoa, you get two doldrums!” was how Mau kidded Nainoa over this spell of light airs where he had been expecting steady trades. On April 11 the wind began

- 70

Hōkūle'a on the way to Tahiti. (Will Kyselka)

- 71

to pick up again, and from then on until Tahiti was reached on April 15 brisk SE trades sped the canoe along its way.

Upon leaving the ITCZ Nainoa estimated that Hōkūle'a was well to the east of his reference course line, and decided that it would be most expeditious to sail directly south for Tahiti. Accordingly, although the canoe could have been pointed further to the east into the wind, Nainoa set a heading of one compass point to the east of south so that, after the effects of leeway and current, the canoe would actually be sailing due south.

This meant that, instead of simply allowing the canoe to sail herself to windward, the steersmen now had to use the stars by night and the sun and swells by day to steer a specific star-compass course. Because many crew members were not skilled at such steering, Nainoa found that he had to stay awake virtually around the clock in order to keep the novice steersmen on course. To ease that burden he called on Milton Bertlemann, an expert steersman who had sailed in Hōkūle'a to Tahiti in 1976, to take over the supervision of the steersmen when he needed to sleep.

When Nainoa estimated that the canoe was east of the reference course upon leaving the ITCZ, he was essentially correct, although he placed it some 30 miles farther east of where it actually was at the time. That gap between the estimated and the actual position tripled between sunset on March 30 and midnight March 31 when the canoe's course veered sharply to the south-west, apparently because Hōkūle'a encountered a narrow zone of current moving swiftly to the west. Recent oceanographic research has shown that near the equator, where the Coriolis effect is small, the current may temporarily reach two knots or more. Analysis of the canoe's track, heading and speed indicates that the current was strongest between 3 °N and 1 °50 ′N. With light easterly winds, the canoe sailed across that latitude at two to three knots on a heading one point (11.25 °) east of south, while her track was three points west of south. This divergence of approximately 45 ° between heading and track is consistent with a current speed of two knots, some four times faster than the average current speed Nainoa was using in his dead reckoning calculations. Although he had been briefed about the possibility of such an accelerated current flow, Nainoa had no way of detecting it at sea. Hence, when the canoe's course was sharply curving to the south-west, Nainoa continued to estimate the course as straight south. Accordingly, on April 1 the canoe was some 90 miles west of his estimated position (Figure 10).

For the next 10 days Nainoa was able to estimate the course of the

- 72

FIGURE 10.

The 1980 voyage of Hōkūle'a (equator to 10°S) showing actual track and Nainoa Thompson's D.R. positions. The number beside each D.R. position refers to the date: “a” for estimates made before noon, “b” for estimates made after noon. The dashed line joins the D.R. position with the true position at the time.

canoe fairly well, which meant that the same gap of around 90 miles persisted as the canoe sailed southward. However, the slow sailing during this period, particularly in the “second doldrums”, made Nainoa worry that he was, in effect, underestimating the distance to the west the canoe had been set by the South Equatorial Current. Because in his dead reckoning system current drift was built into his reference course line, and because the canoe had taken six days more to cross this stretch of ocean than he had allowed for in drawing up his reference course, he - 73 reasoned that he would have to add in the extra six days of current to bring his reckoning in line with reality. So, as the wind began to pick up on the 11th, Nainoa moved his estimate of the canoe's position almost 2° (120 miles) to the west.

In making this bold revision Nainoa had added in some extra miles to the west just in case he might be overestimating the canoe's windward performance. Remarkably, the resultant D.R. position of April 11 was only 32 miles to the south-west of the canoe's true position. Nainoa had no way of knowing this, however, so the tension over the navigation experiment continued to mount as the canoe sailed on (Figure 11).

The sighting of a black noddy tern skimming over the waves on the same day he revised his position estimate did not serve to set Nainoa's mind at ease. This small bird, one of the prime species used for landfinding, ordinarily flys out to sea only a short distance from its island home. Yet, here was one of them skimming over the waves at a time when Nainoa reckoned that they were still hundreds of miles from the Tuamotus. He was sure that he could not be that far off in his latitude estimate, but he began to worry that the canoe might have made more distance to the east than he had calculated and be near the Marquesas Islands, or, conversely, that it might have been pushed far enough to the west to be approaching tiny Caroline Atoll. Only when Mau dismissed the bird sighting with words to the effect that it meant nothing did Nainoa relax, remembering that during his training Mau had told him that groups of birds feeding at sea, not single birds flying alone, were the sure indicators that land was near. Nainoa concluded that he should ignore this lone noddy tern and trust his star sights and his dead reckoning.

During an interview tape-recorded two days later, in the pre-dawn hours of April 13, Nainoa estimated that Hōkūle'a would reach 14 °S sometime during the daylight hours. Since the most northerly of the Tuamotu atolls lie less than one degree (60 miles) farther south, this meant that they should see land soon: “So tomorrow”, Nainoa said, referring to the daylight hours of the 13th, “look for birds . . . and for sure look for them in the afternoon. Maybe we'll see land tomorrow afternoon or tomorrow night; surely by . . . the following morning or day.”

Accordingly, at dawn a close watch was mounted to look for birds or any other sign of land. As the day wore on, anxious eyes spotted a number of noddies and white fairy terns, some flying singly, others in groups. Everyone, including Mau, accepted this as a sign that land must be near. But no coconut palms poked their leafy crowns above the horizon, and at dusk the birds did not oblige their watchers by grouping

- 74

FIGURE 11.

The 1980 voyage of Hōkūle'a from Tahiti to Hawaii showing actual track, reference course and Nainoa Thompson's D.R. positions. The number beside each D.R. position refers to the date: “a” for estimates made before noon; “b” for estimates made after noon. The dashed line joins the D.R. position with the true position of the canoe at the time.

together and then flying in a straight line back to their island home.

In fact, although Nainoa's D.R. estimate at this time was less than 20 miles west of the canoe's actual track, it was almost 70 miles too far to the south (Figure 11). Judging from latitude estimates he had made previously when the skies were as clear as they were now, Nainoa should have been able to estimate his latitude to within about a half a degree, 30 miles or so. Perhaps a vision that haunts navigators when approaching - 75 the Tuamotus—that of the unyielding reefs fringing each atoll of what used to be called the “Dangerous Archipelago”—led Nainoa unconsciously to overestimate the southward progress of the canoe so that there would be no possibility of being caught unprepared by a reef suddenly appearing ahead. At any rate, Nainoa's overestimate of latitude meant an anxiety-filled night during which everyone was up staring at the horizon searching for land. Dawn broke without a sighting, although birds could be seen in all directions, darting among the waves and diving to catch their daily fish. Finally, at mid-morning, first coconut palms, then the surf-bordered line of an atoll came into view. Hōkūle'a had reached the Tuamotus on April 14, the 30th day at sea, having made the passage two days faster than she had done in 1976.

Nainoa had made it to the Tuamotus as planned; all that remained was the short sail to Tahiti, but the question remained as to which atoll lay ahead. Nainoa thought that the canoe had intersected the chain somewhere along its western end, but was not sure exactly which island it could be. As the canoe sailed close enough to gauge the size of the island, both Mau and Bertlemann, the only persons on board who had sailed in the canoe to Tahiti in 1976, declared that it was too large to be Mataiva, the westernmost atoll of the chain where Hōkūle'a had made her landfall on that first voyage. The suspense was broken when the Ishka told them that the island was Tikehau, the atoll immediately to the east of Mataiva.

The landfall on Tikehau effectively ended the navigation experiment. In ancient times a navigator experienced in those waters would probably have recognised the island, or if it were unfamiliar to him he could have asked the inhabitants the identity of their island. Similarly, the direction from the island to Tahiti would either have been known to the navigator, or could have been obtained from the island's inhabitants. Secure in the knowledge that their position was fixed and that Tahiti was an easy day-and-a-half's sail away, Nainoa broke off his long concentration and lay down for his first real sleep of the voyage. The burden of navigation was then given over to the escort vessel.

During that night, however, the Ishka inadvertently took a course about 20° east of the correct heading. Although Nainoa could tell this from a glance at the star field, he let the escort vessel keep its lead. When no land appeared on the horizon the next day, Nainoa, who well knew that Tahiti lay off to the west of them, gave the order to change course towards land. That afternoon gathering clouds and rain-showers reduced visibility, but just as it was growing dark crew-member Jo-Anne Kāhānāmoko Sterling spotted the outlines of mountains, and soon afterwards all could see the flashes of the Point Venus lighthouse at Tahiti's - 76 northernmost extremity.

This incident explains the dog-leg at the southern end of the track line in Fig. 11, and underlines both the value of the invariable star compass and Nainoa's skill in using it.

Where the voyage to Tahiti is largely one long, hard struggle to gain eastward distance against the trades, because of Tahiti's windward position the return to Hawaii can be an easier and faster sail. In 1976 Hōkūle'a had made it back to Hawaii in only 22 days, much of which time the canoe was sailing somewhat off the wind. Before the 1980 voyage Nainoa had calculated a reference course line by first plotting a theoretical course sailed hard into the wind. Even with conservative calculations of the canoe's windward performance, and ample allowance for current drift, the resultant line intersected the latitude of Hawaii at a point over 300 miles east of the archipelago. This margin meant that the canoe could afford to sail off the wind, gaining speed, and still reach Hawaiian waters at a point safely to windward of the chain.

However, although the alignment of Tahiti and Hawaii with respect to the trades might make for a relatively fast sail back to Hawaii, it does not make the navigation any easier. With the freedom to sail swiftly off the wind comes the necessity of keeping the canoe on the right star course. Instead of passively monitoring the canoe's course when sailing hard into the wind, as was done for much of the way to Tahiti, Nainoa now had to pick out the proper course and make sure the steersmen kept to it. This Nainoa had anticipated, for he had several expert steersmen join the crew in Tahiti. None the less, setting and holding the right course remained a formidable challenge, as did the nerve-racking job of mentally plotting the canoe's progress, and above all of making sure that when Hōkūle'a reached Hawaiian waters she would be to windward, not to leeward, of the island of Hawai'i.

In addition, Nainoa was concerned that the Hawaiian Islands, especially when approached from the windward, presented a somewhat smaller target than had the wide arc of atolls and high islands he aimed for on the way to Tahiti.

Hōkūle'a left Pepeete late on the afternoon of May 13, after a layover of almost a month in Tahitian waters. Hopes were high for steady SE - 77 trades to carry the canoe swiftly to the north-north-east. However, visions of fast sailing off the wind under fair, trade wind skies soon succumbed to the reality of the next 10 days: variable winds, including strong breezes which sometimes blew from slightly north of east; towering cumulus clouds and high cirrus formations that often hid the stars; and spells of drizzle and soaking rain. Not till May 23 did the expected SE trades male their appearance, only to fade into a drizzly calm the next day, and then to come back for a few more days to speed the canoe northward at a pace of six knots or so.

Despite the frequent wind shifts that made it difficult to keep an accurate mental picture of the changing course of the canoe, and despite the many nights when clouds hid key stars, Nainoa was able to keep the canoe on the desired heading, and also to estimate its course with some accuracy (Figure 12). His D.R. positions shadow the actual track of the canoe, although slightly to the east of it, and only for a few days did his latitude estimates differ much from the canoe's actual latitude. These slight differences between his mental picture of the canoe's course and its actual track did not progressively increase as the canoe sailed farther and farther to the north. On the 28th, after two weeks at sea, Nainoa's D.R. position was within 21 miles of the true position.

By then Nainoa was a much more confident navigator than when he started out for Tahiti. He felt that he had proved himself, and was fully qualified to take the canoe back to Hawaii. When Hōkūle'a had arrived in Tahiti Mau had taken the first opportunity to fly back to Micronesia to see his family. Before leaving he had told Nainoa that he might not be able to get back to Tahiti for the return voyage, adding that he really did not need to sail back because Nainoa would be able to take the canoe back by himself. “If I come back,” he told his Hawaiian colleague, “I come back for the trip—not to help you”.

Mau did come back, for he is not one to miss a good trip. He sailed more as a passenger out to enjoy a good canoe cruise, however, than as a master navigator worrying about his pupil's performance. Throughout the voyage to Tahiti Mau had kept close track of the canoe's course and speed so that he could step in and take over if Nainoa became lost or made a major error. This meant that Mau slept little, particularly during the first half of the voyage when he spent almost all the hours of the day and night watching over the canoe and its progress. On the return voyage, however, Mau relaxed. He began sleeping soundly at night, and even spent several afternoons idly flying home-made kites—all of which would have been unthinkable had he actually been navigating.

- 78

FIGURE 12.

The 1980 voyage of Hōkūle'a from Tahiti to Hawaii showing actual track, reference course and Nainoa Thompson's D.R. positions. The number beside each D.R. position refers to the date: “a” for estimates made at sunrise; “b” for estimates made at sunset. The dotted lines join the D.R. positions with the true position of the canoe at the time. Approximate current boundaries from aircraft expendable bathythermograph observations made along 153°W, May 13, 1980 (Stroup et al., 1981).

- 79

The return crew was lucky; instead of getting stuck for days in doldrum conditions, they found enough wind north of the equator to be able to sail from the SE trades into the NE trades virtually without a hitch. Only between May 28 and 30 did the easterly winds sometimes falter, and squalls, heavy rain and periods of calm hinder the northward progress of the canoe.

The track of Hōkūle'a and Nainoa's parallel D.R. positions would seem to indicate that Nainoa had no trouble guiding the canoe, both in fact and in mind, through this transitional zone. However, on one occasion Nainoa did become briefly disoriented. The canoe ran into a line of squalls during the daylight hours of May 28, after which Nainoa was wary of assuming that the wind was coming from the east as it had before the squalls had struck. The sun was not visible through the heavy cloud cover, and Nainoa could make no sense of the conflicting swells left in the wake of the squalls. Worried that in the absence of clear indicators he might head off in the wrong direction, Nainoa was about to order the sails triced up when, instead, he decided to consult Mau. Nainoa did so not out of desperation—for he knew that eventually the sun or stars would become visible and allow him to reorient himself—but because he wanted to take this opportunity to have Mau further his navigational education by pointing out how to discern directional cues from such confused seas. Mau was able to orient himself from the swells, but he was not able to show Nainoa exactly how he did it. Nainoa would need more experience, Mau told him, before he would be able to read such confused seas well enough to orient himself without error.

At about 8°N Hōkūle'a encountered brisk trades which stayed with the canoe all the way to Hawaii. Because he was confident that the canoe was still hundreds of miles east of the longitude of the island of Hawai'i, Nainoa then gave the order to steer a course slightly to the west of north which, he calculated, should bring the canoe within range of Hawai'i, but slightly to windward of it. It took only six days to reach Hawaiian waters, a fast passage made possible by the angle as well as the strength of the wind. The east-north-east direction of the trades meant that the canoe could be sailed on a reach across the wind, a favourable sailing angle which allowed her to make between six and seven and a half knots for much of the way.

Once in the steady trades, Nainoa re-calculated his position, primarily to take into account the faster than expected passage through the - 80 ITCZ. Before leaving, Nainoa had expected that they would be forced to spend five days in the ITCZ, and that—assuming the wind and current zones coincided—the eastward-flowing Equatorial Countercurrent would give the canoe a significant boost to the east. He had built that expectation of eastward drift into the angle of his reference course line. Now, however, Nainoa decided that since the canoe had actually spent only two days transiting this zone, he had to adjust his D.R. position to the west.

It was a shrewd re-calculation. At sunrise on May 30, before making the adjustment, Nainoa had placed the canoe some 50 miles east of where it actually was. At sunset of the same day, after the adjustment, Nainoa had moved the canoe's estimated position west to a point which in longitude was within five miles of the true position (though somewhat to the south because of an overestimate of latitude).

On June 1 a white fairy tern was spotted. This sighting bothered Nainoa, for at the time he reckoned (correctly, as it turned out) that the canoe was still hundreds of miles from the island of Hawai'i, the nearest land. This was a solitary bird, however, and it acted strangely for such a normally diffident species by flying over the canoe and fluttering off the stern. Mau looked momentarily puzzled, but said nothing at the time; that evening he called Nainoa over and told him that the bird must be sick and lost.

Two days later, on June 3, another fairy tern was spotted flying with a group of tropic birds, a sighting which Nainoa took as a possible confirmation of his calculation that they were nearing the island of Hawai'i. He was reasonably certain of his star sights which indicated that at sunset they should be at about the latitude of the southern tip of the island, and he thought that he had been successful in guiding the canoe on a course which would put it comfortably off the eastern shore of Hawai'i.

Nainoa had indeed kept the canoe on the desired course, for it was then on a heading which would take it past the eastern side of Hawai'i just as planned. However, in his dead reckoning Nainoa placed the canoe's course about 60 miles to the west of its actual track, in effect estimating that the canoe would pass much closer to the island than it actually would. This overestimate of the proximity of land would seem to parallel the time towards the end of the voyage to Tahiti when Nainoa reckoned that the canoe was closing in on the Tuamotus faster than it actually was, and similarly may have sprung from an unconscious desire to make sure that there was no chance of coming across land unexpectedly.

That night Nainoa reckoned that they must be abeam of the island; to - 81 make sure of his latitude estimate he had the canoe lie to with sails triced up, so that he could measure the height of Polaris as precisely as possible. Satisfied that they had truly reached the latitude of Hawai'i, Nainoa ordered the canoe under way again.

The next morning, with no island yet in sight, Nainoa ordered a course change. Because he reckoned that continuing on their present course would soon take the canoe north of Hawai'i, he set a new course directly to the west, where his calculations told him the island was located.

But the low clouds and haze typical of strong trades made it difficult to see very far ahead. Complicating matters was the difficulty the crew experienced in holding the canoe on a westerly course following a break in the centrally mounted steering sweep ordinarily used for downwind sailing. That forced them to tack downwind, steering the canoe with side-mounted steering paddles instead of sailing directly before the wind.

That afternoon Nainoa was wakened from a brief nap by a question from Mau: was he certain that the downwind turn was correct? Mau had not questioned Nainoa at any time earlier in the voyage but (as Nainoa was intently aware) this turn was the critical point of the northbound leg. If a gross error in navigation had been made, so that the canoe were actually west of the islands rather than east, the turn would be taking them away from land and putting them in an extremely difficult position far to leeward of their goal.

With the possibility of this very serious error always in mind, Nainoa had been careful to check his navigation at every stage, but Mau's question prompted him to review his decision once more. He went forward to be alone, and started to go over all his mental calculations. As he did so, however, he intuitively realised that he was indeed correct: the canoe was east of Hawai'i, and there was no need to try to repeat mentally every twist and turn the canoe had taken in the course of the voyage. When Nainoa went aft to tell Mau that he was certain the islands were ahead, Mau instantly accepted his judgment: “O.K., we go that way!” Encouraged by Mau's absolute confidence, Nainoa returned to the search for land.

The glare from the afternoon sun, however, made it virtually impossible to see anything ahead. Not till the sun was setting was the silhouette of one of Hawai'i's 13,000-foot mountains finally seen. Stephen Somsen tape-recorded these comments right after that first glimpse of home:

Nainoa earlier in the day had predicted that one way of seeing land would be to watch the sun set behind the island. So, as sunset approached, he - 82 spent his time in the bow. Just as the sun was coming down, I moved forward and stood in front of the mast. He was on the port side. And as the sun moved towards the horizon, it was eclipsed by one of the Hawaiian islands, probably the “Big Island”. Nainoa just turned to me, extended his hand. We shook hands. Beautiful moving experience. I complimented him, told him he had done a beautiful job and he really has.

The remainder of the voyage was uneventful. Hōkūle'a sailed past the north tip of Hawai'i, along the northern shores of Maui and Moloka'i, then between Moloka'i and O'ahu and on to Honolulu, where she entered the channel leading to the Ala Wai yacht harbour at midday on June 6, after 26 days at sea.

Any scientific experiment should be replicable. Other investigators following the same basic experimental design should be able to obtain essentially the same results. Particularly where controversy surrounds a novel experiment, and the validity of the findings are questioned, a repeat is called for.

The navigation experiment conducted on the 1976 voyage of Hōkūle'a to Tahiti—in which Mau Piailug successfully guided the canoe without instruments—was marred by incidents in which vaguely worded navigation information from the outside reached the canoe (but not Mau), leading some critics to reject the validity of the experiment (Prince 1976). 6 There were even unfounded accusations, which started circulating in Tahiti well before the arrival of Hōkūle'a that there must have been a compass hidden on board the canoe. It would be humanly impossible, the accusers charged, to navigate so accurately for such a great distance without the use of instruments. 7

The 1980 experiment should serve to lay these doubts and accusations to rest. There were no incidents in which navigational information from the outside reached the canoe before land was sighted, and there were no accusations that hidden navigational instruments had been secretly consulted. There can be no question that Nainoa navigated the canoe using solely the information gained by observing the heavens, watching the wind and swells and looking for the appearance of land-based birds.

But Nainoa did more than simply replicate and confirm Mau's feat. He extended the original experiment by navigating the canoe back to Hawaii, again without instruments. The return might seem to have been a navigationally easier task because of the better weather and more - 83 favourable winds, and because Nainoa was by then more confident of his own abilities, but in one important respect the voyage back to Hawaii presented a greater navigational challenge than a voyage to Tahiti.

Although the task of navigating to Tahiti is hardly a matter of being carried there by “favourable winds and currents” (Danielsson 1980:44), both Mau (in 1976) and Nainoa (in 1980) made landfall at the western end of the Tuamotu archipelago largely by trimming Hōkūle'a to sail herself to windward, on the port tack, for most of the way. If the same practice had been followed on the return to Hawaii (trimming to sail as close to the wind as practicable on the starboard tack), the canoe would have passed hundreds of miles to windward of the Hawaiian chain. Although Tahiti's windward position relative to Hawaii may allow a returning canoe to be sailed swiftly off the wind for much of the voyage, it also meant that Nainoa had to set the canoe on the right star compass course for each segment of the voyage, and make sure the steersmen kept to those courses. The Hawaiian navigator did his job well; he passed the ultimate test of landfall by homing in on the island of Hawai'i just as he had planned. In so doing he demonstrated that his skills, and by extension non-instrument navigation methods in general, are up to the task of sailing over thousands of miles of open ocean where wind conditions require constant attention to accurate course setting and steering.

The 1980 voyage also provides an interesting perspective on another objection to the way in which the 1976 experiment was conducted. In 1976 (as in 1980) Hōkūle'a was shadowed by a yacht for purposes of both documentation and safety. Heyerdahl (1981:38), Prince (1976:8), and other critics, 8 assuming that the canoe was actively steered all the way to Tahiti, have contended that the steersmen could not help being influenced in their choice of heading by glancing at the heading of the yacht trailing astern. The implausibility of this suggestion is indicated by the fact that Hōkūle'a was trimmed to sail automatically to windward for much of the 1976 voyage, and that the escorting yacht was forced (because of its superior speed and pointing ability) to sail alternately to one side or the other of the canoe's track in order to keep from overtaking Hōkūle'a and disappearing to windward. The incident that occurred during the 1980 voyage just after the canoe cleared the Tuamotus and briefly gave the lead to Ishka for the final run for Tahiti underscores this implausibility, for it shows that following an escort vessel would not necessarily be more accurate than following the appropriate horizon stars.

Because Nainoa's system includes some non-traditional elements, his accomplishment cannot be considered as an exact duplicate of the way - 84 traditional Polynesian navigators might have once guided a canoe to Tahiti and return. Yet, it is those non-traditional elements—such as his method for determining latitude in degrees, and his way of expressing the canoe's deviation from the reference course—which make Nainoa's effort so useful for examining the issue of long-distance navigation in prehistoric Polynesia. These elements have allowed us to plot his D.R. positions on a chart and thus make a precise comparison of his mental picture of the canoe's progress with its actual track; this in turn enables us to examine the issue of the accuracy of non-instrument navigation from a uniquely informed perspective.

Akerblom (1968), Hilder (1959, 1963), Sharp (1956, 1961) and others skeptical that Pacific Islanders could have ever made purposefully navigated voyages over long distances stress the problems inherent in non-instrument navigation. To begin with, they claim that using horizon stars to set and hold a course is inherently inaccurate, both because it is impossible to align a canoe exactly on a desired star course and because the bearing of horizon stars shifts slightly as one sails north or south. Course error is increased significantly, they further claim, when clouds obscure the stars, and during the day when there is only the sun, wind and swells for orientation. Storms, variable winds and shifting currents must, in their conception, inevitably compound error. Even when clear skies and steady winds are re-established after a storm or period of variable conditions, they doubt that a navigator could get back on the proper course, because he would have no way of knowing where the canoe had been pushed by wind and current. All this adds up, they say, to the impossibility of long-distance navigation without instruments. Sharp, the most vociferous of the skeptics, even went so far as to proclaim a maximum sailing distance of 300 miles, beyond which purposeful navigation was in his view impossible.

The record of the 1980 voyage, particularly that for the leg from Hawaii to Tahiti, contains much information that could be taken as confirmation that there are major sources for error inherent in a non-instrument navigation system. For example, on the way to Tahiti Nainoa did find it difficult to gauge the canoe's heading exactly, particularly during the many cloudy days and nights experienced on that leg of the voyage. The stormy conditions encountered soon after leaving Hilo were disorienting, as were the many cloudy days in the ITCZ. The canoe was also subject to unexpected and undetected current drift such as the strong westerly “jet” encountered in 3°-2°N. Despite these problems, Nainoa was able to guide the canoe to its planned landfall on the Tuamotu chain. Furthermore, he was able to cope with a somewhat dif- - 85 ferent set of navigational problems on the return, and to bring Hōkūle'a home without difficulty. How can we explain why a navigation system which would seem to have so much room for error should prove to be so workable in practice—particularly in the hands of a navigator making his first long voyage?

To begin with, we suggest that the skeptics have overestimated some of the difficulties of non-instrument navigation over long distances. Take, for example, the way the problem of long-distance navigation is typically posed: to find a small island after sailing over hundreds of miles of open ocean. As a glance at the map of the Pacific will show, most of the islands are located in long archipelagic chains; it is the entire chain, not the individual island within, which furnishes the navigator with a reasonably attainable goal. For instance, the extensive Tuamotu chain to the north and east of Tahiti, and to a lesser extent the long Hawaiian chain, provided broad targets for Nainoa.

Then, to move from the question of island targets to the sea itself, consider the common argument that, because the non-instrument navigator cannot perceive changes in current flow, the cumulative effect of those changes must inevitably cause the navigator's dead reckoning to stray farther and farther from reality.

According to this reasoning, the unusually strong current flow encountered on the way to Tahiti from 3°N to 2°N should have thrown Nainoa's dead reckoning well off to one side of the actual course, and any subsequent current shifts should have widened the gap. Indeed, because Nainoa could not see the current jet north of the equator, and therefore could not have realised that it was carrying the canoe westward at a considerable rate, at this time his dead reckoning positions did begin to diverge sharply from the canoe's actual track. But, after 10 days of being significantly off in his dead reckoning, at 11°S Nainoa moved his D.R. positions almost directly on to the track of the canoe. Yet, he did so for the wrong reason. Nainoa figured that, because of the slow passage through the unexpectedly light airs encountered south of the equator, the canoe was being subjected to a more westerly current set than he had built into his dead reckoning framework. He therefore calculated that he had to move his dead reckoning to the west to reflect this added current set. In fact, the current south of the equator was unusually weak at this time, as indicated by the approximate parallelism of Nainoa's D.R. positions and the actual track of the canoe (Figure 10). The overestimate of current set south of the equator thus cancelled out the underestimate of current set north of the equator.

Although the happy outcome of this sequence of current shifts and - 86 mis-estimates may be dismissed as pure luck, we propose that it illustrates an important feature of long-distance, non-instrument navigation. While it might seem common sense to assume that longer voyages will inherently be more difficult because navigational “errors” will tend to accumulate and progressively throw the navigator farther and farther off course, this case, and similar ones discussed by Lewis (1972:222-3), would seem to argue for the reverse. Navigational “errors”, whether arising direct from unperceived environmental causes or simply from misjudgments of star bearings or wind directions, do not necessarily accumulate in one direction; they are more likely to be random and hence will tend to cancel each other out as the voyage lengthens. 9

Yet, neither an appreciation of the breadth of the targets formed by island chains, nor an appeal to randomness, fully explains why Nainoa did not become completely disoriented, then lost, in the immensity of the Pacific. A considerable degree of skill is also involved. Both legs of the voyage are replete with instances where Nainoa demonstrated his navigational ability. For example, look at what happened on the way to Tahiti when the weather finally cleared after leaving Hilo, and again when the canoe left the ITCZ and entered the zone of the SE trades. In both cases, after being significantly off in his dead reckoning because of cloudy skies and irregular winds, Nainoa was able to re-establish where the canoe was with considerable accuracy. Furthermore, towards the end of both legs, after sailing over thousands of miles of open ocean, Nainoa was able to put the canoe on just the right bearing towards land. In addition to exaggerating the problems of non-instrument navigation, it would appear that the skeptics have also underestimated the skill of the navigators.

In attempting to explain precisely how Nainoa navigated, there is an almost automatic analytical bias towards a formalistic presentation, stressing the principles applied at each stage of the navigational process, and the sequence of particular methods employed. This is, for example, virtually built into Lewis' (1972) description of traditional navigation and the summary given in this paper. In some respects, it is even apparent in the way Mau instructed Nainoa during their formal training sessions. Furthermore, in his effort to get it right the first time, Nainoa certainly attempted to apply each procedure with systematic rigour. Yet, as Will Kyselka brings out in his forthcoming book on the 1980 voyage, An Ocean in Mind, Nainoa's navigational technique is made up of more than just the sum of the individual observations and calculations; there is also an important intuitive element. There were times when Nainoa seemed to know where the canoe was, and which course to set, without - 87 consciously going through all the observations and calculations he had so carefully learned or developed.

In criticising what he considers a false, formalistic approach to human thinking pursued by advocates of Artificial Intelligence, Hubert Dreyfus stresses the importance of intuition—“everyday non-mystical intuition”—which he defines as “a learned response based on past experience” (quoted in Rose 1985:51). In a study of how fighter pilots and chess players acquire and exercise their skills, Dreyfus has found that only novices follow formal rules. As they become more proficient, people depend more and more on context and experience. Expert fighter pilots fly by the feel of it, rather than by the book, and chess masters, unlike their computer challengers, do not analyse hundreds of moves; they sense the right one when it pops into their head. What guides these experts is not so much step-by-step analytical thought, but intuitive response based on experience.

Nainoa's experience as a navigator was certainly minimal when Hōkūle'a left Hilo, but as the voyage progressed he was able to start making the subconscious integrations of observation, experience, and training—the intuitive leaps—which are so difficult to explain analytically.

In addition to being aware of the differences between some of his methods and those of Mau, Nainoa was sensitive to the contrast in style between him and his teacher. Nainoa felt that, where he tried to take a step-by-step approach to navigation, Mau seemed to take a much more holistic approach. In terms of the Dreyfus study, this may reflect the difference between the beginner and the expert, or perhaps, more accurately—given Nainoa's remarkable first voyage and his flashes of intuition—the difference between an aspiring navigator in the process of becoming an expert, and a superbly experienced master.

Traditional navigation has essentially disappeared in Polynesia and in much of Micronesia. Mau is even pessimistic about its survival on Satawal. “Maybe I'm the last navigator,” he says, for none of the younger Satawalese seem to have the time or the dedication to learn enough to become fully qualified master navigators. Nainoa is one of the few Pacific Islanders today trying to master non-instrumental navigation. Yet, despite the success of the 1980 voyage, Nainoa is the first to admit that he still needs to learn much more. Although he has acquired a high proficiency in setting and keeping track of his course by the stars at - 88 night and by the sun, wind and swells during the day, Nainoa still has difficulty in reading direction from swells where conflicting seas obscure regular patterns, in using disruptions in swell patterns to ascertain proximity to land, and in orienting himself on dark, cloudy nights when the wind is frequently shifting in direction. Also, Nainoa has yet to master the etak dead reckoning system used by Mau, relying instead on his own methods.

An extended voyage scheduled for 1985-1987 in which Mau and Nainoa will navigate Hōkūle'a from Hawaii to New Zealand, via Tahiti and the Cook Islands, then to Tonga, Fiji and Samoa, and finally back to Eastern Polynesia and home to Hawaii via the Marquesas, will give Nainoa an opportunity to further develop his navigational skills. In addition, other aspiring navigators, including Mau's youngest son, will sail on various legs of the voyage in order to obtain first-hand instruction and experience in traditional navigation.

Although the resulting enhancement of Nainoa's navigational skills and the training of other Island navigators during the voyage are not likely to add up to anything like a full-scale revival of this vanishing art, the cultural impact of this effort in re-learning that began in 1976 should not be underestimated. Voyaging canoes, the artefact which made the settlement of the islands of the Pacific possible, evoke in Islanders a proud vision of their ancestors migrating into the vast reaches of the Pacific to find and settle virgin lands. Those who today can navigate voyaging canoes without resort to compass or other modern aids personify this vision; they are a living link between present and past.

In this cultural sense the 1980 voyage went far beyond the successful replication of an experiment. The 1976 voyage of Hōkūle'a has been hailed as one of the events that opened the current “Hawaiian Renaissance” (Kanahele 1982), and also had a great impact upon the Tahitians and upon other Islanders who heard about it. Cultural considerations have been part of the project from its inception. The Hōkūle'a venture was organised in 1973 as more than just an effort to test the sailing ability of the Polynesian double-canoe and the accuracy of non-instrument navigation methods. The organisers also hoped to promote cultural revival among Hawaiians and other Polynesians by having them take principal roles in the building, sailing and navigation of Hōkūle'a. In 1976 this joining of two objectives, one cultural and one experimental, proved much more difficult than expected (Finney 1979). In 1980, however, the two came together smoothly, with cultural revival enhancing research and vice versa. Not only was a significant experiment carried out, but also, for the first time in centuries, a Polynesian - 89 navigator followed the stars, winds and swells over the long sea road between Hawaii and Tahiti. 10

• Åkerblom, Kjell, 1968. Astronomy and Navigation in Polynesia and Micronesia. Stockholm, Ethnografiska Museet.
• Alkire, William, 1970. Systems of Measurement on Woleai Atoll, Caroline Islands. Anthropos, 65:1-73.

• Danielsson, Bengt, 1980. Hokule'a: In Pursuit of an Unlucky Star. Pacific Islands Monthly, 51(12):43-44.

• Finney, Ben R., 1967. New Perspectives on Polynesian Voyaging, in G. A. Highland, et al. (eds), Polynesian Culture History. Honolulu, Bernice P. Bishop Museum Special Publication No.65, pp.141-66.
• —— 1977. Voyaging Canoes and the Settlement of Polynesia. Science, 196(4296):1277-85.

- 90

• —— 1979. Hokule'a: The Way to Tahiti. New York, Dodd, Mead.

• Gladwin, Thomas, 1970. East is a Big Bird. Cambridge, Harvard University Press.
• Goodenough, Ward H., 1953. Native Astronomy in the Central Carolines. Philadelphia, The University Museum, University of Pennsylvania.

• Heyerdahl, Thor, 1981. With Stars and Waves in the Pacific. Archaeoastronomy, 4(1):32-8.
• Hilder, Brett, 1959. Polynesian Navigational Stars. Journal of the Institute of Navigation. 12:90-7.
• —— 1963. Primitive Navigation in the Pacific-II, in Jack Golson (ed.), Polynesian Navigation. Wellington, The Polynesian Society, pp.81-97.

• Kanahele, George, 1982. Hawaiian Renaissance. Honolulu, Project Waiaha.
• Kyselka, Will, in press. An Ocean in Mind. Honolulu, University Press of Hawaii.

• Le Journal de Tahiti. June 16, 1976, Papeete, Tahiti.
• Les Nouvelles. June 17, 1976, Papeete, Tahiti.
• Lewis, David, 1966. Stars of the Sea Road. Journal of the Polynesian Society, 75:85-94.
• —— 1972. We, the Navigators. Honolulu, University Press of Hawaii.
• —— 1977. Mau Piailug's Navigation of Hōkūle'a from Hawaii to Tahiti. Topics in Culture Learning (5):1-23.

• Prince, A1, 1976. Hōkūle'a Does it But it Wasn't “Pure Polynesian” say Critics. Pacific Islands Monthly, 47(8):8-10.

• Rose, Frank, 1985. The Black Knight of AI. Science 85, 6(2):46-51.

• Sharp, Andrew, 1956. Ancient Voyagers in the Pacific. Wellington, The Polynesian Society.
• —— 1961. Polynesian Navigation to Distant Islands. Journal of the Polynesian Society, 70:219-26.
• Stroup, Edward D., Bernard J. Kilonsky, and Klaus Wyrtki, 1981. AXBT Observations during the Hawaii/Tahiti Shuttle Experiment. Hawaii Institute of Geophysics Report HIG-81-1.

• The Tahiti Bulletin, June 18, 1976, Papeete, Tahiti.

• Wyrtki, Klaus, et al. 1981. The Hawaii to Tahiti Shuttle Experiment. Science, 211(4484):22-8.