Volume 89 1980 > Volume 89, No. 4 > Etak and the ghost islands of the Carolines, by Michael J. Gunn, p 499-507
                                                                                             Previous | Next   

- 499

Within the orally transmitted systems of navigational knowledge recorded from Puluwat by Riesenberg (1972), there are many phenomena listed that are hard to reconcile with the techniques used in oceanic sailing and orientation. These phenomena include the two “ghost” or “vanishing” islands of Kaafiror and Fanuankuwel, 1 as well as varied items of “sea-life”—a category that includes such diverse items as two-headed whales, groups of umwole fish, and frigate birds. The problem under consideration in this paper is: what are the ghost islands and sea-life doing in the oral systems of navigational knowledge? The ghost islands may operate as a communication device, connecting the composite “map” of the world conveyed in the systems of knowledge to the world as it is sailed. The sea-life in one of the systems of knowledge will be shown to have a role in developing the navigational construct of etak used during all voyages.

Carolinian navigation has been systematically studied since the major German expedition of 1908-1910, in which the three islands of Puluwat, Satawal, and Woleai were found to have flourishing schools of navigation. As a consequence of this expedition, Sarfert (1911) and Damm and Sarfert (1935) gave comprehensive accounts of the principles of etak navigation on Puluwat, the sidereal or star compass, and the Notinseln or ‘emergency islands’ used in etak navigation. They suggested that etak was a method of measuring the relative distance between islands, and that it used a third island—the Notinsel—moving backwards under the horizon stars used in the sidereal compass.

The techniques of actually sailing the big ocean-going canoes (waa) were recorded by Thomas Gladwin (1970) and by David Lewis (1972), and it is from these two accounts that we have learned of the perceptual acuity of the Puluwatan palu ‘navigators’. It appears that training a palu begins more or less informally when he first sails on the waa of his relatives, goes through formal initiation, and culminates some 10 to 20 years later when he has mastered the techniques of sailing, orientation, weather prediction, and the associated taboos and chants. A large part of the palu's training involves the honing, clarification, and shaping of perceptual techniques. Using no external system of measurement, apart from the magnetic compass (a recent introduction) to maintain direction - 500 when visual clues are obscured, the palu has to rely on his senses aided by a body of interpretive theory.

Eleven major systems of navigational knowledge were collected from Puluwat by Saul Riesenberg (1972). Consisting of lists of islands, sea-life, and other phenomena to be found along the bearing lines that radiate out from each island in their navigable world, this body of knowledge is orally transmitted, and forms the interpretive aspect of Carolinian navigation. It is a portrayal of a world organised around three basic problems. The first is the need to know in which direction a destination island can be found in relation to the point of origin. To negotiate successfully the seaway between two islands, the palu needs to know two further things: how to continually relate himself to both his point of origin and to his destination in terms of direction so that, in the event of being blown or having drifted off his course, he does not miss his destination. A less urgent, but just as important a problem, is in organising the relationship between islands in the region normally sailed to determine the quickest or the safest routes.

Per Hage (1978) has attempted an analysis of the listing structure of some of these systems of knowledge. He saw the problem in terms of the techniques of oral transmission and attributed the regularities he found to a form of mnemonic device. However, many of the regularities in the systems of knowledge can be seen to have far more then mnemonic function. Primarily, navigational knowledge must enable people to locate themselves with reference to certain aspects of the environment known to be constant. Mnemonic devices are undoubtedly there, but they would serve a supportive rather than primary function.

To pursue this line of reasoning further it is necessary to note that the three problems associated with organised perceptual data into navigational techniques generate a need for the palu to abstract his data from the environment. Abstraction has the advantage of allowing the data to be manipulated and synthesised with theory. In Carolinian navigation the world is ordered through both selective and constructive perception. In other words, what is needed to be seen is seen, and if it is visually obscured or non-existent then it is imagined. A simple example is found in their construction of star-paths that cross the heavens, connecting the stars into a series of 16 east-west bands. Given a sighting of any one star plus knowledge of his orientation to the heavens that he can deduce from swell patterns, the palu mentally reproduces the star-paths as well as the associated sidereal compass.

Given this basis, we can now consider the aspects of the environment that the palu have treated as constant points in their systems of navigational knowledge. We can elucidate three groupings on the basis of verifiability:

First, we have the islands, islets, and atolls that can be identified as environmental points from at least 20 miles away—10 miles as a minimum sighting distance in daylight, and a further ten miles for the range of predictable land-based bird-life at sunset or sunrise (Lewis 1972: 126, 162-4; Gladwin 1970: 196).

The second category would include reefs and banks that are discernible only when one is immediately above them. They can be spotted from above by the colour change of the water, provided that they are no deeper than 20 or 30 fathoms (Lewis 1972: 212, 231); or at night by “fishing” with a weighted line - 501 (Gladwin 1970: 58). Identification of these points is usually complete if the confirmed position was predicted by the sailing instructions for the voyage, but if the navigator was unsure of his position, identification would come on recognition of the geomorphology or the associated sea-life.

Those points in the environment that are stable, constant, yet intangible should be considered as the third grouping of environmental constants. They include the points of intersection between two or more seaways, islands outside the detection radius, and the two “ghost” or “vanishing” islands positioned north of the Carolines between the archipelago and the Marianas.

Using the sidereal compass, the environmental constants outlined above are used through bearing “lines” radiating out from each point and connecting to other recognised points. The palu learns the relationship between each point so that Truk, for example, is said to lie under the rising of Altair (due east) when viewed from Puluwat. Between the constant points of the environment lie the exigencies of the less constant or even downright inconstant aspects of the open ocean. For these the Carolinians have honed their sailing skills to an admirable art (see Gladwin 1970; Lewis 1972).

Etak (or hatag as it is termed in Woleai) appears to be an integral aspect of Carolinian navigation. It is used on even the shortest journeys. For example, Alkire records that the voyage from Woleai to Ifaluk is said to be three hatag long:

“. . . the first is counted when Woleai drops from view astern, the second covers that part of the voyage which lies between this point and the point when Ifaluk first appears on the horizon, and the third and final one lies between this last point and Ifaluk itself” (1970: 54).

Etak is a technique for the imaginative reconstruction of the environment, in which topography—real or contrived—is represented as a focus for inverting the image of the sidereal stars along part of the horizon. The inverted image is generated within the palu by his knowledge of the approximate location of the etak island, and controlled by his awareness of his rate of movement in relation to the fixed aspects of the world. Etak is a method of measuring the gaps between islands; the inverted image functions as a device for converting and bundling “rate of movement” into discrete units (see Fig. 1).

Distance is important, particularly on long voyages when the navigator may be required to turn in towards his target when approximately parallel to it. The major problem of handling distance comes not with the actual estimate of rate through the water, but in the cumulative logging of distance travelled. A palu may be at sea for a week at a time, continually estimating rate of travel. If he had no way of converting rate into proportion of journey travelled, he would be left with no means of knowing his east-west position.

The Carolinian navigator, then, is left with the problem of finding some standard by which he can fix his position on the sea's surface. Time could be used, and is used for short voyages of a day or so, but for longer voyages time measurement becomes increasingly impractical. For, as Lewis pointed out (1972: 120), the palu Hipour and Tevake were less able to state the time of night or day than he, whose own estimations were from 10 minutes to an hour or more out. The

- 502
Basic Principles of Etak.

number of days could be counted in terms of dawns, dusks, noons, but this information is flexible, depending on the currents, winds, and other phenomena affecting the speed of the canoe, and hence is not transposable in accurate terms. The palu has to use rate as well.

As Gladwin noted (1970: 186-188) etak segments will often be of unequal length, but as the palu is fully aware that the angles between the points of his sidereal compass are also unequal, this problem is not unsurmountable. He also notes, however, that on exceptionally long voyages an etak island may be selected so as to give relatively even etak segments, and that these segments may be so far apart that a full day's sailing may be needed to cover just one. Lewis's voyage with Hipour, from Saipan to Pikelot, used Gaferut as an etak island, and three days had passed before Gaferut was alongside (Lewis 1971: 444). After reference to Hipour's representation of wowfanu given in Riesenberg (1972: 30) in which he has listed both Kaferoor and Fanuankuwel on either side of the seaway between Guam and Gaferut, one is inclined to suspect that Hipour mentioned only islands that both he and Lewis knew were there. Furthermore, Gladwin noted that the two navigational traditions of Faanur and Warieng differ on the number of etak islands they would use on a long voyage, and that these differences in the selection of etak islands are a major point of distinction between the two schools (1970: 200-201). Navigators of the Warieng school, Hipour among them, often select two islands with a shift from one to the other midway through the voyage.

- 503

The suggestion that things are grouped “. . . not to facilitate action, but to advance understanding, to make intelligible the relations which exist between things” (Durkheim and Mauss 1963: 81) throws light on the role the ghost islands play in the navigable world. In the body of recitations that carry a large part of the palu's knowledge of navigation (see Riesenberg 1972), all three classes of environmental constant are linked in orientational terms. On this level the palu's navigational knowledge is a body of accretions acting as a “map” of the environment, for a map in some way mirrors the world in the terms through which the map is organised. In the Carolinian setting the connection between the map and the world usually coincides with an island, or bank, or reef. These points are constant and during a voyage are used, through etak, to calculate location when out of sight of land. However, in the 500km gap between the Carolines and the Marianas Group there are no identifiable physical landmarks to connect the body of knowledge to the environment. In this instance an island is needed to bridge the gap so that navigational requirements are satisfied. It is in this position that the two ghost islands are said to be located: Kaferoor to the west of the seaway running from the Central Carolines to Guam, and Fanuankuwel to the east. 2

The navigational system “The Fortune Telling of the Sea Bass's Food” recorded by Riesenberg (1972) from Tawuweru, gives us an example of etak operating within the “map” context of the navigational knowledge. This system appears to

Puluwat as Etak Island for Lulunna Shoal
- 504

portray one of the initial training runs given a palu for developing his etak concept. The directions given are obviously not the sailing directions for the normal safe run from island to island. To start with, the journey from Pulusuk to Puluwat is usually sailed under Star 1 (due north) with Lulunna Shoal used as an etak reference island on one side of the seaway, and with Truk as another etak island on the eastern side in the system used by the Faanur school (Gladwin 1970: 188). Tawuweru is of the Faanur school, and yet in this system he describes the palu as sailing under Star 30 up Manila Reef to the end of Lulunna Shoal, at which point he turns and sails under Star 4 to Puluwat (see Fig. 2).

The deviation from the normal course to Puluwat is three sidereal star positions, from Star 1 to Star 4. In the recited course from Pulusuk to the end of Lulunna Shoal there are given two intermediate phenomena, both identifiable banks:

From Pulusuk under Star 30, to:

  • 1. Pukuelailai (Long Bent Knee), a bank, so called from its shape, in the southern part of Manila Reef; thence to:
  • 2. Apilepil (Heel), another bank named from its shape, this being the northern half of Manila Reef; thence to:
  • 3. Lul (Breaker), the Lulunna of the map, so named because of the waves breaking over this reef. Thence: under Star 4, to Puluwat (taken from Riesenberg 1972: 34).

The journey is then completed to Puluwat whereupon the same course heading and the same number of phenomena are cited, taking the sailor in position to the southern tip of Gray Feather Bank—due north (under Star 1) of Lulunna Shoal (see Fig. 3).

- 505
Truk as Etak Island for Ruat to Ulul

The journey would cover two etak marks if Lulunna Shoal was used as the reference island. If Ulul was the reference island there would be only one marker, about midpoint. The recitation gives two markers, Raikaniang and Leeniwor, indicating that Lulunna Shoal is used as a reference island and that the student navigator would sail the same physical distance with the same (approximately) spaced etak markers, but no physical markers. The physical position of Gray Feather Bank is a mark indicating the turning point and providing confirmation of the third etak position. With this stage completed, the palu then sails to Ulul.

The recitation continues after Puluwat:


Under Star 30, to:

  • 1. Raikaniang (Fish of Iang, which is the name of the space of water between Puluwat and Ruat); this is a whale of the same species and appearance as the second whale of VIII Aligning the Skids, but smaller, about the size of a paddling canoe and with the upper fluke missing; thence to:
  • 2. Leeniwor (Reef Pond), so called because it consists of a school of bonito churning about in a circle and producing frothy water like the water running over a shoal; thence to:
  • 3. Ruat, which is apparently southern Gray Feather Bank (see III, the Sail of Limahacha). Thence:
- 506

Under Star 4 to:

  • 1. Eeylipon (pon: murky water), where a shark lives, visible only occasionally because of the cloudiness of the water; thence to:
  • 2. Merupul (pul: flame colour), a bank, identified by a flame coloured fish, four feet long, of species mwenitaw; thence to:
  • 3. Ulul; thence to:
  • 4. Magur.

Under Star 14 to Piseras.

Under Star 13 to Truk.

Under Star 22, return to Pulusuk (taken from Riesenberg 1972: 34-5).

The course from Gray Feather Bank to Ulul would cover three etak if Truk was used as the reference island. Truk lies under Star 10 from the supposed position of Ruat, and lies under Star 13 from Ulul. In the recitation there are two phenomena, Eeylipon and Merupul, to cover before the palu reaches Ulul; and these correspond in number to the etak shifts a palu would use if sailing this route (see Fig. 4).

This analysis of the “Sea Bass's Food” indicates that the phenomena listed between constant environmental points may well operate in a purely functional way. In this instance they operate as etak markers, very possibly as a training run to develop the etak concept used as a rate marker.

The point to note is that in this training run the palu is as perceptually blind to the etak islands of Puluwat, Lulunna, and Truk as he would be to an imaginary ghost island. An etak island does not need to exist in physical terms. It only need operate as a constant point in the ocean with its position fixed relative to other constant points in order to function as a connection between the “map” of navigational knowledge and the physical world.

It is in this sense that the two ghost islands of Kaferoor and Fanuankuwel exist in the ocean north of the Carolines. Connected to the other constant points by sidereal bearings, they fill the gap that exists in the physical world. Without these two islands the ocean north of the Carolines would be completely empty for the 560km from Gaferut to Guam. The palu's image of where he was would have nothing to support him for at least three days, yet he would still be sailing within the area covered by his learned body of knowledge. The web of bearing lines that connect island to island within the archipelago continually reassures the palu when he crosses them; he is secure within the bounds of his knowledge. But when travelling north of this web, the bearings connecting him with the other islands of the Carolines would be open-ended, with the canoe suspended between the archipelago and the open ocean. We can say, then, that the two ghost islands operate as etak reference points, providing the palu with a means of processing rate and at the same time uniting his physical position on the ocean to his navigational knowledge.

- 507
  • ALKIRE, William H., 1970. “Systems of Measurement on Woleai Atoll, Caroline Islands.” Anthropos, 75: 1-73.
  • DAMM, H. & E. SARFERT, 1935. in G. Thilenius (ed.), Ergebnisse der Südsee-Expedition 1908-1910, II, B, 6/2, Hamburg.
  • DURKHEIM, E. & M. MAUSS, [1903] 1963. Primitive Classification, Cohen and West, London.
  • GLADWIN, Thomas, 1970. East is a Big Bird. Harvard University Press, Cambridge, Mass.
  • GOODENOUGH, Ward, (1953), Native Astronomy in the Central Carolines, Philadelphia.
  • HAGE, Per, (1978), Speculations on Puluwatese Mnemonic Structure, Oceania XLIX (2) pp. 81-95, Dec. 1978.
  • LEWIS, David, (1971) A Return Visit Between Puluwat and Saipan Using Micronesian Navigational Techniques, Journal of the Polynesian Society 80(4), pp 441-447.
  • —— (1972) We, the Navigators, A.H. & A.W. Reed, Wellington.
  • RIESENBERG, Saul H. (1972), The Organization of Navigational Knowledge on Puluwat, Journal of the Polynesian Society, 81(1), Mar 1972, pp 19-56.
  • —— (1975) The Ghost Islands of the Carolines, Micronesica, 11(1): 7-33, July 1975.
  • SARFERT, E., (1911), Zur Kenntnis der Schiffahrtskunde der Karoliner, Korrespondenzblatt der deutschen Gessellschaft fur Anthropologie, Ethnologie, und Urgeschichte, 42.
1   Kaferoor has a strong existence in secular mythology from Puluwat as well as from islands in the western Carolines. Its mysterious property of vanishing from sight when approached caused three islands to the south-east of Kaferoor to be misnamed by Western cartographers (Riesenberg 1975).
2   The existence of the two ghost islands in the navigational knowledge may have come about through an hypothesis that land existed in these locations. These “lands” may well have been used as etak reference islands in voyages between the Carolines and the Marianas islands. As these voyages were successful, Kaferoor and Fanuankuwel can be seen as unconfirmed but useful hypotheses.