A Case for determining that Ouro and Luca.Antara Islands shown on the Eredia Map are, respectively, Melville and Bathurst Islands of the Tiwi Islands of Australia

   This Website has been established to allow a wider readership of the paper by Noel H.Peters, published in the December 2003 issue of CARTOGRAPHY, the  journal of the Mapping Sciences Institute, Australia.
Readers are encouraged to study the paper and the Triple Connection and Gridlock technique for comparing an antique map to its modern counterpart to decide for themselves if it is more likely than not that the Eredia Map 1602 depicts the Tiwi Islands of northern Australia


Note: For ease of comparison and printing, Plates 2 - 6 below are also available in improved, electronic form - see separate link.




A Case for determining that Ouro and Luca.Antara Islands shown on the Eredia Map are, respectively, Melville and Bathurst Islands of the Tiwi Islands of Australia


N.H. Peters


Europeans, the Dutch, first discovered Australia in 1606.  This paper claims there was probably an earlier Portuguese exploration of the Tiwi Islands in northern Australia around 1600.  And it puts forward a novel bearing technique for comparing antique maps to corresponding modern publications. The technique is used to demonstrate that a 1602 map, by Emanuel Godinho de Eredia, then a leading cartographer, probably depicts these islands.  Some background to Eredia and his map, plus the political background necessitating the survey, is outlined along with an assessment of the identity of the likely surveyor. Finally, detailed cartographical and topographical results are presented with a summary indicating their significance.


It is generally accepted Europeans first discovered Australia in 1606, and that a small part of its coastline was surveyed, at Cape York Peninsula, Queensland, during the voyage of the Dutch ship Duyfken. (Heeres, 1899; McIntyre, 1977; Ward, 1987). The map by Emanuel Godinho de Eredia, dated 1602 (Plate 1), depicts in its southern half two islands, Ouro and Luca.Antara. The purpose of this paper is to present the hypothesis that these two islands are the Tiwi Islands of Australia, namely Melville and Bathurst Islands respectively, and to demonstrate this.  The paper also asserts the map of each island provides corroborative evidence of the other, and that they demonstrate a Portuguese survey and charting of sections of Australia before that of the Dutch.


The subject of this research is the antique Eredia Map (1602) that was only discovered in 1946 at the National Library in Rio de Janeiro by Dr. Mota Alves (McIntyre, 1977)It represents a cartographical detective story and provides evidence, to a high degree of probability, that sometime around 1600 the Portuguese made a detailed coastal survey of the two main Tiwi Islands which they named Ouro (Melville Island) and Luca.Antara (Bathurst Island).

Emanuel Godinho de Eredia, to whom the map is attributed, was in his day probably the leading cartographer, cosmographer, geographer and mathematician of East Asia, from Goa in the west to the Spice Islands, and north to China and Japan. Spate (1957) considered In     matters of survey and cartography,...Eredia is worthy of considerable respect. He was the son of a Portuguese father and a Malayan princess, and educated by the Jesuits first at Malacca and then at Goa, the Portuguese enclave in India (Mills, 1930). His maps of these parts are well documented (Cortesão and da Mota, 1960).  Eredia also wrote a book, The Description of Malaca, with the manuscript dated 1613 (Eredia, 1613).  In it, along with fifty-six maps and drawings, he wrote a small chapter on Meridional India (South Indies) containing a detailed report on an island called Luca.Antara to which a journey was made in 1601.

Eredia specifically leaves his reader to infer that he did not visit these islands, yet he was careful to add his name to his 1602 map; and to another world map, dated 1606, on which he names himself as the descobridor of Luca.Antara in 1601 (Cortesão and da Mota, 1960). This particular Luca.Antara (1606), however, was a peninsula on a great tract of land in the south. So what and where did Eredia really intend by the name Luca.Antara? And was this particular 1606 inscription an intentional ruse by Eredia to mislead readers away from a proper knowledge of newly discovered islands in the south, and away from the nearby coast of mainland Australia? Descobridor in Portuguese means discoverer. Spate (1957) followed exactly the Mills' interpretation of descobridor and the Commission granted Eredia in 1594 by Philip I of Portugal (also Philip II of Spain). He repeated Mills' view that the Commission gave Eredia the priority rights to discover it at some future time. Mills (1930) says: Descobridor, that is, officer commissioned to organize the work of exploration and discovery.

As will become apparent, there is good reason to believe that Eredia, in 1601, was prepared to manipulate the truth regarding his discoveries in the south. This paper maintains he propagated a hoax that the King of Damuth, Chiaymasiouro (unknown) had visited an island called Lucaantara (sic). And Eredia claimed the evidence of this was the very descriptive letter written by Chiaymasiouro to the King of Pam (also unknown). This letter published by Eredia details the journey, the island's inhabitants, their King's hospitality, and the produce of the country including gold (Eredia, 1613). His deception is dealt with immediately before the conclusion to this paper. In essence, Eredia's writings, some twelve years after the event, appear to be a continued, deliberate smokescreen aimed at confusion by deception, subterfuge and secrecy for the protection of Portugal's long-term, political interests. One can more easily understand Eredia's deception when it is realized that from 1595 ...and by 1601, no fewer than 14 Dutch fleets (at 3 to 5 ships per fleet) had sailed for the Indies and the United Provinces had overtaken Portugal as the leading trading nation in the East (Dash, 2002).  Against this, it is important to note that Eredia's main work was naval activity, and that he had placed at his disposal the whole southern squadron of some seventy boats (Mills, 1930). 

Commercial and Political Background

To create a united front in their commercial assault on the Spice Islands, the Dutch formed the Verenigde Oost-indische Compagnie, the VOC, commonly called the Dutch East India Company, in 1602, the same date as this Eredia map. Dutch competition was creating intense pressure on Portugal in the East Indies, challenging for the Spice Islands trade. This was for the cloves from five islands off the west coast of Halmahera as well as mace and nutmeg from the Banda Islands, due north of the Tiwi Islands (Napier, et al, 1973).

Over seven short years, 1595 to 1602, the Portuguese were fast losing their grip on vast sections of their eastern empire, not just in Ambon.  Apart from the Dutch, the Malays were constantly on the attack around Malacca (Mills, 1930). The Ambonese were in revolt, and the English, by 1601, had also formed a trading company that despatched five ships to Java (Napier, et al, 1973). In the end, the Dutch took permanent control of Fort Victoria and Ambon Island in 1605 (Muller, 1993).  In short, the region was in utter political turmoil.  Yet in the midst of this, clearly the Portuguese were surveying the Tiwi Islands.  Perhaps they considered these islands as a possible alternative base.  As Eredia reveals, the Portuguese also had a fort at Ende on the island of Flores (Eredia, 1613).  There was a real motive for them to be quietly looking at other territory to the south around 1600-1601, and Eredia finished this map in 1602.


The reader is asked to judge the evidence presented by answering the simple question:  Is it more likely than not that the maps of Ouro and Antara (Plates 1, 3 and 5) depict Melville and Bathurst Islands, Australia (Plates 2, 4 and 6) and provide clear evidence that the mapping was carried out prior to 1606.  The more likely than not test is an alternative accepted interpretation of the balance of probabilities test of evidence applied as constituting proof in all civil matters throughout the legal systems of the English speaking world.


It is submitted that the Eredia Map (1602) is two maps in one. It depicts a simple, general map in which the island of Ouro is shown as a place to and from where a sea journey was made, arriving from Ende and departing to Timor, or vice-versa.  But it also shows a very detailed portolan map of the possible arrival area (Luca.Antara) to be found at the southeast end of a direction line from Genteng, South West Java. The island marked Ouro, with IAP written off the east coast, is somewhat large when compared to known islands to the north, if Ouro were to be Melville Island (Melville).  The island marked Luca.Antara (Antara) is almost as large as Java Major (Java), and if it were to be Bathurst Island (Bathurst) (Plate 2), then it is greatly enlarged. The two islands of Ouro and Antara are shown south of Java, Bima (Soembawa), Ende (Flores) and Timor. These last four islands are roughly in proportion to one another.

Longitudinal error is considerable if these islands are to be Bathurst and Melville.  But latitudinal error is small if Genteng or nearby Ratu Point, South West Java is taken as the starting point for the only rhumb line connecting Java to the south west of Antara (Plate1).  In fact, the latitude of Genteng is in error by 2° 47' compared to its correct latitude.  Ouro and Antara have been placed far apart, and on a first impression, may appear to have little connection to anything that may be part of the Australian mainland or islands somewhere off its coast.  McCaskill (1996, pers.comm.) having meticulously studied this Eredia map material, however, noted.....the general similarity in island shapes to those of the modern map. They are no more distorted than those of Java Major (Java), Bima (Soembawa), Ende (Flores) or Timor, the outlines of which had no doubt been derived by Eredia from the surveys of earlier 16th century Portuguese navigators in The Indies.



Plate 2


 Plate 3



By contrast, Spate (1957), McIntyre (1977) and Ward (1987) have all suggested as a possible solution that Ouro and Antara are, in fact, part of the Australian mainland. All re-drew the Eredia map as being part of the mainland, and Spate adorned the inside back and front covers of his book Let me Enjoy with his novel cartography. While Spate considered the two islands to be imaginary, McIntyre and Ward even nominated the landing place, Collier's Bay - Brunswick Bay in Western Australia, and confirmed it in writing.  By comparison, the author has merely enlarged the two islands by two grades on a photocopier. This produced images of Ouro and Antara approximately the same size, respectively, as Melville and Bathurst on a modern map.  No pretence whatsoever is made regarding the change in scale required to achieve this equality in size. This is not essential to a bearing map, discussed below, nor could the mariners and surveyors then measure coastal distance accurately. Bearings, therefore, influenced shape and not size, and in terms of orientation, the mapping convention of a North pointer had not yet come into general practice by 1600.  Toponymy does not form part of the case being presented.

Bearing Maps

In the context of this research, a bearing map is one where the placement of salient coastal features and the shape of an island under survey are dictated by specific bearings measured along the alignment of two or more prominent objects along the coast or in the hinterland.  The following examples illustrate the technique.  One bearing line was obviously taken along the coast of Ouro (Plates 2, 4 and cf: 3). See Cape Van Diemen to Brown Point, and compare CVD to BP; and then see Cape Van Diemen - Y - (probably Hill 44) - Gordon Point, and compare CVD - Y - GP.  Bearings were also taken across an island, from a coastal point to an inland point of elevation and then on to another salient coastal point (Plate 2. See X - (probably near 410) - Notch Peak 100 - Cape Gambier. (Cf: Plates 3 and 4). The first bearing could have been taken from X, rising to near spot point 410, and then down to Notch Peak 100 (a very prominent landmark) and on down to Cape Gambier. The accuracy of any such bearing is limited by the prevailing topography, and the availability of adequately positioned elevation to facilitate sighting and estimation of distance. Other physical impediments would have included rivers, and terrain, and perhaps hostile inhabitants. These bearings and the sketching of map segments appear to have been supplemented by observation and estimation of both coastal shape and distance between salient points.  Estimation of distance also had to take into account the position of the reciprocal salient point on the relevant bearing.

Outstanding examples of similar alignments were found, with many probably used to construct the bearing map, on Antara/Bathurst (Plates 5 and 6).  There are eleven bearings across Mission Hill (Peak 93), and twenty bearings across the centre of the island at Peak 102 / X102.  Then there were also nine bearings (only two TCs are shown at One Tree Point (OTP) in Plates 5 and 6) radiating from Penguin Hill 81, Bowen Bay on the south coast (far west).  The Peak X102 bearings were discovered as a result of the inland elevation concept. There are several obvious hinterland features marked on the Ouro and Antara maps (Plates 2, 5, and cf: 6 then 4).  Whilst internal detail is limited, these are features worthy of investigation. The reader should note on Ouro (Plate 1) there is a clearly marked black dot in the mid-north sector just north of the 'ro' in the word Ouro.  This will be shown to be Peak 68 on Melville. And on Antara, there is a clearly marked large black dot shown in the extreme eastern sector. This will be demonstrated to be Mission Hill (Peak 93) (Plate 1), the third highest point on Bathurst Island. There is also a broken, curving line marked clearly, east of the word VEL. and west of the only opening on the mid-east coast (Plate 1). It indicates a rock face and contour (Contour 50) clearly visible to the west from the largest opening on the northwest coast of Antara (Euro Creek). And a river or location was often enlarged for importance and easy recognition.  This largest opening (Euro Creek)(with the large black dot immediately to the east) indicates a safe harbour, Euro Creek, 700 metres wide at the mouth and just eleven miles northwest of the southern entrance to Apsley Strait from Beagle Gulf. Moantu, on the south coast (west), was also enlarged. There is an abundant supply of fresh water at Moantu, and water then was a regular, major problem for all mariners.



 Plate 4


For the purpose of comparison of Ouro with Melville Island, the map of the island of Ouro has been enlarged off the Eredia map (Plate 1) with a photocopier.  Enlargements have been made to produce a copy (Plate 3) approximately the same size as the modern map of Melville (Plate 2).  A general similarity between Melville and Ouro, and between Bathurst and Antara, (Plates 2 and 3) becomes apparent.  The comparison between the original and modern maps as well as their subsequent identification have come about from a process involving a study of general shape. The shape comparison relies upon a novel 'Triple Connection' (TC) Technique discussed below, and the resultant triangles and trapezoids formed.  All of these are used in comparing coastal features as well as the grids, unique to each island, thereby created.  Another vital factor is the inescapable conclusion, from this and earlier research (Peters, 1992), that inundations, mud and sand flats, and saline coastal flats were treated as sea in the Eredia map, and not land as in modern maps.

Triple Connection Technique

The Triple Connection (TC) Technique detailed in Table 1 is a method used in order to compare, by bearings, the shapes of landmasses represented on maps of different or unknown scale, and to demonstrate that a landmass in an antique map is the same landmass depicted in a corresponding modern map.

A straight line that joins three or more points or locations is termed a Triple Connection (TC).   A location may be simply a point on part of a coastline where the line crosses or intersects it or a fixed point in the landscape. The Triple Connection formed by CVD  -  near T  -  X  -  BP is one example shown on Plates 2, 3 and 4.  A true bearing can be established on the modern map for all points along the TC, but for the Eredia map it is sufficient to conclude that all points along the TC have the same bearing. The novel Triple Connection (TCs) Technique was developed as part of earlier research by the author (Peters, 1992).

Highlighting the triangles and trapezoids formed by the Triple Connections (Plates 2 and 3) was a logical, subsequent development for the purpose of comparison of the maps through evaluating the similarity in shape of these geometric figures.  In the earlier research by Peters, Ouro was identified by a limited number of TCs, as was Antara.  Further research has been conducted to create a more extensive network of TCs.  Ouro has now had its unique triangles and trapezoids, created by the TCs, clearly shown for the purpose of comparison. There are nine TCs on Ouro/Melville (Plates 2 and 3) and fifty-one TCs on Antara/Bathurst that have been identified.

The TC Technique is important because the concept of scale and the ability to measure distance accurately were not well developed in 1600.  Therefore, a comparison of general shape plus the shape of a coastline and its features, by using bearings, is considered to be more relevant and reliable than assessments based on size or scale.  An informed comment has been made (McCaskill, 1996, pers.comm.) that ...others will need to devote a little time to work through your cogent cartographic arguments and to appreciate not only the difficulties of coastal charting in the sixteenth century, but also the extraordinary care and accuracy with which cartographers transcribed the surveys on to published maps.  It is also necessary to accept your very reasonable arguments that direction, measured by compass bearing, is more important than distance...

Using the TC technique, salient points on Ouro have been identified against corresponding points on Melville, and vice-versa, and then examined for comparison.  In coming to an understanding of the maps of both Ouro and Antara, and in interpreting the impact of their TCs plus the shape of the resultant triangles and trapezoids, it is also critical to remember that the maps are bearing maps.  As such, bearings and direction were more important than general shape or size. Shape was subservient to bearings and direction.  And direction took preference over distance.

Points identified on the old maps have mostly been allocated coded reference letters (for example, CVD for Cape Van Diemen, CG for Cape Gambier and so forth) and the elevation of a hill is noted in the label such as Peak 68, where applicable.  These are marked on Plates 2, 3 and 4.  Examples of a bearing connecting relevant TC points on Ouro are:

CVD  (Cape Van Diemen)  -  Y  (St. Asaph Bay indentation)  -  GP (Gordon Point);

CVD  -  near T (East of small coastal indentation and mouth of creek)  -  X  ( a point in Shark Bay south of Purumpenelli Point)  -   BP (Brown Point)


Gridlock is achieved with the intersection of two TCs.  Gridlock can occur anywhere along the bearing of each of the TCs and may be at an identified point common to the TCs or simply a point of intersection (Figure 1). This term, coined by the author, is used to describe the locking of the unique grids formed by each set of points defining the TCs.  Once gridlock is established, the points define a unique spatial arrangement that is invariant. This is despite any changes in scale that may be applied and caused by variations in the ability to accurately measure distance at the time. Gridlock, with the other TC points, locks in and fixes the relationship between the points.  Triangles and trapezoids can be created as a result of this relationship, if relevant, merely by joining two adjacent extremity points of a gridlocked pair of TCs  (See also CG  -  CP for the five-sided figure marked by the symbol of a Star on Plate 3). The arrangement of the points involved in the TC of the gridlock, and the invariant shapes of the triangles and trapezoids, may be compared between maps at different proportion in order to assess the similarity of the detail contained within the maps.

Distortion and Rotation

The map of Ouro is obviously distorted when it is compared to a modern map. Distortion resulted, firstly, from errors in measuring bearings. It cannot be said, of course, which of the correct bearings were actually taken by the surveyors at that time, but it is clear they also made errors in, or failed to survey, other bearings. This results in the need sometimes to rotate the old map in different places to achieve alignment of comparable bearings and features. Secondly, as shown by Plates 1 - 6, distortion also occurred because measurement of distance between salient points was probably random and inconsistent.  Lack of technology ensured that direction, and simple estimation of distance and coastal shape became paramount.  Having determined one segment of the coastline, the surveyor was assuredly confronted with the need to juggle distance in one or more other segments in an attempt to make them all fit the jigsaw puzzle of bearings.  This was either in an adjoining segment or in the segment where the reciprocal point of a given bearing was located. If the distance errors were more likely random than systematic, then this caused a distortion of shape as the errors manifested themselves and mounted across the growing map segments.

There is an excellent example of making the segments fit the jigsaw puzzle on the southeast coast of Antara/Bathurst, immediately west of Taradiri Creek, marked "G" on the south coast of Bathurst.  Note G to B (Plate 6 and cf: Plate 5), and the indentation on the coastline of Antara that runs southwest not west. This involves some four nautical miles of coastline. On the Antara map, the same distance is shown approximately as more than double that of the modern map. Here was an enigma where no identification could be made or comparison found.  The solution was discovered finally on the author's second field study of the area. On examination, it became obvious the coastline had been stretched substantially.  This was done, along with mapping the coastal indentations in correct sequence and approximate proportion, to fit the jigsaw puzzle. The convention or habit of treating coastal inundations and sand flats then as sea, not coastline, had been adopted. (The explanatory maps and photographs are not presented here). This accounted for the old, different coastline shape compared to the modern equivalent. In other words, one segment was distorted out of proportion to another, but whether this was intentional or otherwise would be speculation. Shape, however, was the casualty as errors in distance are more likely to distort proportion. Distance is not fixed, only direction.


 Plate 5


     Plate 6


For an amplification of the following Figure 1, see Figures 1 - 4 in the subsequent paper Eredia, Ouro and Luca.Antara, and Nova Guinea on this website.

Figure 1



When making a comparison, it may be necessary to rotate the old map and the TC bearing in order to reach alignment with the TC of the modern map.  Rotation has long been a feature of understanding antique maps.  This is partly due to the absence of a convention to indicate North or the incorrect alignment of a map segment.  In any event, when the old map is rotated, as necessary, one can achieve a more satisfactory result. Another explanation for rotation may be the necessity to take up an original, inaccurate estimate of coastal distance. Alternatively, it could be to account for an original, intentional adjustment of distance for the surveyor to accommodate other salient points on given bearings (which may have been inaccurate anyway). This probably means the reciprocal point of a bearing is within a different map segment that is out of proportion.

Rotation, then, is the simple, random rotation of the old map caused by the inaccuracies in measurement or placement of the bearings. This may be accounted for at various points by rotating the old map around a point so as to align a TC bearing on the old map with a corresponding modern TC. This removes rotation as a factor in the matching process.  Correctly defined and aligned TCs, if made in situ and at the time, may generally require little or no rotation (See the gridlock at CVD - Plates 2, 3 and 4). Yet this is not always the case. Using two rulers, testing a super-accurate TC, with four connections running across the width of Antara/Bathurst (Plates 5 and 6), demonstrates this.  Compare: SEC (South East Corner) - Peak 93 (Mission Hill gridlock) - Z (Euro Creek) - DC (Dudwell Creek, northwest coast). The rotational error is 24º. Again, in these old bearing maps: Bearings and direction were more important than general shape and size. And direction took preference over distance.

Corrected Error

The Corrected Error technique, developed by the author, relies upon the comparison of the latitudes of salient points within the old map extent, particularly Ouro, by assuming a corrected latitude of a well defined point established from a modern map of known accuracy.  The Corrected Error technique has been applied in this instance by measuring the latitude of a known place, Genteng or nearby Ratu Point on South West Java, shown on the Eredia map. This is then compared to its modern, correct latitude to find the difference or error, in this instance 2° 47'. This same error is then applied to the latitude of salient points on the island under examination to provide an improved estimate of the latitude of features on the Eredia map.  But a similar measurement taken along, say, the south coast of Antara would be irrelevant because Antara was intentionally mapped out of all proportion and size to Ouro. (See RESULTS below - paragraph 2).


Marginalia is the variable margin of parts of an old map's coastline where it is used to show from the sea the topography along and near that coastline.  The author has previously written on marginalia, a term he coined then, demonstrating their use on old maps and how they should be interpreted (Peters, 1992).  All marginalia appeared to be different in some respects, yet there is a common thread running through the old methods.  Whilst there is a detailed marginalia on the Antara map (the explanatory maps are not presented here), there is no evidence of marginalia on the Ouro map. The accuracy of the marginalia along the south coast of Antara could only have been the work of a very competent, experienced surveyor (Eredia was one). This marginalia may well represent the results of a monocular view of coastal parts where distance may have been extremely difficult to estimate, yet it still demonstrates a very careful, physical examination and charting of the coastal topography. McCaskill (1996, pers. Comm.) examined the particular marginalia technique used on Antara and concluded: The numerous topographical identifications along the south coast of Antara/Bathurst Island. What's impressive is the correct sequence of high and low ground.


Comparing the old and modern maps and the unique grid of TCs for each island, with and without rotation, shows a remarkable correlation between them.  The evidence indicates that, irrespective of the distortion of both islands when compared to a modern map, a substantial difference in proportion, and despite the necessity for rotation to achieve alignment as required, the islands of Ouro and Antara are considered to be Melville and Bathurst respectively when the Eredia map is subjected to the more likely than not test of evidence mentioned at the outset.

By applying the Corrected Error technique, it was found that the northernmost point of Ouro, marked CVD (Cape Van Diemen), is in error by only three nautical miles, at 11°10' South.  And the southernmost point, marked CG (Cape Gambier), was in error by a mere seven nautical miles at 11° 58' South.  By applying the same technique to Antara, it was found that the northernmost point, marked BP (Brace Point), was in error by just ten nautical miles at 11° 19' South. It would appear the errors in latitude were systematic not random.

The large black dot on the Ouro map (Plate 3) in its upper one-third sector, now marked "68", is a multi-gridlock position formed by five TCs. This point and its corresponding five TCs on Plate 2 indicate a remarkable similarity when the rotation indicated previously is taken into account. This correspondence on the two maps is further convincing evidence that the maps are, indeed, of the one and the same landmasses.  A further striking feature of the Ouro map is the ability to compare and test other gridlock positions on it, according to the TC and Gridlock techniques in Table 1, by checking the points or locations along an old TC against those on the corresponding modern TC, and by noting the angles formed.  Cape Van Diemen, Brown Point, Cape Gambier, Cape Keith, Gordon Point and Peak 68 are all gridlock points that have corresponding gridlocks on Ouro. Importantly, each gridlock corroborates the others.   In addition, the five-sided figure, at CG - CP, marked by the symbol of a Star on Plates 2 and 3, together with the adjoining very small triangle on its northeast border, is another solid example of corresponding, comparative evidence.

A visual comparison of the numerous triangles and trapezoids is easy to accomplish, and demands a compelling conclusion of total correspondence between the figures constructed in Plates 2 and 3. This is irrespective of the effect of the skewing caused by the small errors inherent in the mapping and measurement technique.  It must be stressed that the triangles and trapezoids are not, and cannot be, identical, but are geometrically similar within the limits of accuracy of the mapping techniques, and that they can be matched notwithstanding.

This research has shown that inundations, mud and sand flats, and saline coastal flats were treated as sea, not land, at the time Manuel Godinho de Eredia constructed his map.  And note again the south east coast of Melville (Plate 2) with its huge inundations and saline coastal flats against the main coastal indentation on Ouro (Plate 3) east of CG (Cape Gambier with Muranapi Point) between CG and CP (Conder Point). It is a particularly important result.  See also the coastal saline flat immediately northeast of CP (Conder Point), and its interpretation on the old map where it is not included as part of the landmass and where the shape of the inundation is specifically indented at and as the salient point of two bearings, one joined to Peak 68 and the other to BP (Brown Point). The shape of CG (Cape Gambier) on both maps is worthy of special note, with the absence of the little Irrititu Island immediately offshore. The omission of small coastal islands was a frequent event.

Finally, as corroborative evidence, Antara can also be shown to be Bathurst.  This is done by applying the same TC technique with and without rotation; with resultant triangles and trapezoids; by analysing a total of fifty one TCs; by examining the marginalia and topography on its south coast; by establishing with the TC technique Antara's large black dot in the far south east section of the island as a multi-gridlock point, and as being Mission Hill (peak 93) (Plate 2), at present the location of a TV repeater station.  Of this large black dot, McCaskill (1996, pers.comm.) had a query: If this were an inky smudge....why did it so randomly mark so accurately one of the few prominences on the Island?  The nearby white spot (marked on the south coast marginalia at G) so accurately marks the minor eminence, Hill 10, that a random blip just here in printing would be extraordinary;  and by not forgetting to examine the broken, curving line west of the huge opening on the mid-east coast of Antara against the distinctive (contour 50) escarpment and rock face west of Bathurst's huge Euro Creek.

No mention has been made in that list of the most important point of all in the centre of Antara, a multi-gridlock point marked X102, with its many gridlocked TCs passing through or near Peak X102.  This gridlock corresponds exactly with Bathurst's second highest point, Peak 102 on the Ranku, from where visibility extends for miles in all directions.  Having investigated it in depth, McCaskill (1996, pers.comm.) found Peak X 102 to be of crucial ("gridlock") importance.... for so many of the lines of triple connection. As an integral part of the many gridlocks formed at a common point, it is central to the comparison of the Eredia and modern maps of the area.

The author has determined these comparisons in a more detailed set of coloured maps covering both Tiwi Islands (Ouro/Melville and Antara/Bathurst).  A coloured set of them, together with an expert's report (McCaskill, 1996, pers.comm.), was donated to the Western Australia Maritime Museum in 1988.  In 1996, the late Murray McCaskill, Emeritus Professor of Geography at Flinders University, Adelaide, prepared this expert professional opinion answering the question:

Are the islands of Ouro and Luca.Antara more likely than not Melville and Bathurst Islands respectively?

His opinion, given after confirming every triple connection on both islands, was based on meticulous study of the more detailed set of coloured maps covering both Tiwi islands.

Who made the physical survey of Ouro and Luca.Antara, Eredia?

Spate (1957) was a substantial student of the Eredia material, yet appears to have overlooked Mills' English translation of Eredia's book plus Mills' detailed notes and opinions (Mills, 1930).  (Mills was later elected to the Council of the Hakluyt Society).  Spate suggested There seemed little doubt .... Eredia had something solid on which to base his persistent claim to be the pre-emptive Discoverer and Magistrate of India Meridional, which is the Isle of Gold ("ouro" means "gold" in Portuguese), the Great Southland, Terra Australis, Australia. Notwithstanding, Spate concluded ... one of the few things unmistakably clear in Eredia's writings is that he never sailed to India Meridional and never claimed to have done so.

Unfortunately, Spate accepted at face value the so-called letter sent by the (unknown) King of Damuth, Chiaymasiouro (hereafter C's letter), to the (unknown) King of Pam, and not to Eredia. C's letter was undated, yet the rest of Eredia's material was carefully dated, detailed and documented.  C's letter, which can only be judged as complete twaddle, just happened to be recovered by Eredia, and it purports to describe a voyage to Lucaantara (sic), the country and its people, the gold fillets in their hair and their addiction to cock-fighting!  In Spate's defence, however, he did class the letter as a weird document indeed!  So, it is important to ask: What cartographical and geographical skills, education and coastal survey experience did the so-called King of Damuth and his sailors possess if they were the ones who produced the coastal survey sketches from which such a spectacular bearing map of Antara-Bathurst was constructed and in such detail? And how did a King of Damuth generate a variable marginalia, a European mapping convention, on Antara's south coast to the extent it can be read off 400 years later?

Likewise, Spate accepted the two (unnecessary) certificates, both dated October 4, 1601, sworn by Pedro de Carvalhaes, the Captain of the Portuguese fort at Ende, Flores. By them, Eredia purports to authenticate C's letter and then seeks to divert attention from the island of gold, by claiming that another proposed voyage, this time by Carvalhaes, never eventuated. Both certificates are also open to question for the ambiguous (similar) manner in which they are sworn. Carvalhaes declared: Since I have been asked for this information by the Descobridor Manuel Godinho de Eredia, in the interests of his voyage (and enterprise), and for the advantage of the King's service: I swear by the Holy Gospels that this is the truth (the whole truth) and that it is my signature which appears below. (Mills, 1930).  In the light of the maps in this paper, it is a reasonable conclusion that Carvalhaes was really saying: It is the truth that Eredia has asked for this information, not that my (Carvalhaes) whole statement is true.

Major (1873), then a leading geographer and Keeper of Maps at the British Museum, created a substantial academic difficulty for himself over a mappe-monde on which Nuca Antara appears with the inscription Nuca Antara for descoberto no anno 1601 por Manoel Godinho de Eredia (Major, 1861). Having claimed a pre-Dutch discovery of Australia by the Portuguese, based on this curious, confused map (McIntyre, 1977) and its inscription, Major subsequently found it necessary to publish an abject retraction (Major, 1873). Unfortunately, he had failed to be aware and had no knowledge of Eredia's 1613 chapter on Luca.Antara in Declaracam de Malaca ... The confusion ruined Major's academic reputation and broke him (McIntyre, 1977).  But in this honest 1873 retraction, Major attacked Eredia's bona fides, and declared The purposelessness and transparent delusiveness of such a letter (C's letter) suggest to us the high probability of its being entirely spurious.

The author does not accept Eredia's hoax. With the cold suspicion of a lawyer armed with a vital piece of damning evidence, the maps in this paper, the author submits it is more likely than not that Eredia was the physical surveyor, as well as the cartographer, of the Tiwi Islands around 1601.


The degree of correspondence between these TCs, triangles and trapezoids, and the unique grids formed on the maps, plus the location of Peak 68 on Ouro, cannot be dismissed as pure coincidence. Nor can it be suggested that the Ouro map is some landform other than Melville Island when the shape and geometry of the grids match so concisely.  The respective grids formed from the Triple Connections are unique to Ouro and Melville Islands. A further major factor to be considered is that the latitude shown for Cape Van Diemen is but three nautical miles in error compared to the modern map, given the corrected error. From further testing of Antara, using the same techniques, a similar conclusion can be reached that the map of Luca.Antara is more likely than not a map of Bathurst Island, with its northernmost point in error by just ten nautical miles. This confirms the contention that the Eredia Map is two maps in one. Together, the two islands of Ouro and Luca.Antara now provide evidence that, more likely than not, they are Melville and Bathurst Islands respectively, and that each corroborates the other.  It is submitted there was a Portuguese discovery of Australia prior to the generally accepted Dutch first discovery in 1606.

As a compliment to the late Murray McCaskill, this paper concludes with his summation of the Eredia map material McCaskill (1996, pers.comm.):

In summary, one comes away from this scrutiny with a very high regard for the standard of coastal charting performed around 1600 by Portuguese or Portuguese-Malay mariners, their only resources presumably being compass, sextant/quadrant and log. ... I am pleased to support your (Peters) hypothesis of a pre Dutch recording of two of Australia's largest offshore islands.

Cortesão, A.F.Z. and da Mota, A.T. (1960), Portugaliae Monumenta Cartographica,Lisbon,
                      Vol. I-VI; Vol. IV, Pl. 411C, (p.53); Pl. 412A (Fol. 34), and Pl. 412I.
Dash, M. (2002), Batavia's Graveyard, London, p.45.


Eredia, M.G. de (1613), Declaracam de Malaca E India Meridional Com O Cathay, M.S     
                      preserved in the Bibliotheque Royale, Brussels, No. 7264, transl. to English by             
                      Mills, J.V. (1930) as Eredia's Description of Malaca and of India Meridional and       
                      Cathay, M.S. 1613, addressed to the King of
Spain, with Mills' Notes, published 
                       in The Royal Asiatic Society Journal, Malayan Branch, Vol.VIII, Pt.1, 1929-30       
                      (Singapore, Sept.1930)
Intro.pp.1-6 and Notes pp.184-199 (State Library of   
                      NSW Call No.DS506A);  also translated to French, under Ruelens, C. Intro. to                
                      Janssens, L. (ed)(Brussels, 1882) - National Library, Canberra, Call  


Heeres, J.E. (1899), The Part Borne by the Dutch in the Discovery of Australia 1606-             
London, Intro. p.V; p.4.


Major, R.H. (1861), On The Discovery of Australia by the Portuguese in 1601,                  
                      Archaeologia, London, Vol. xxxviii, p.437(1873). 

                      Supplementary Fact relating to the Discovery of Australia, 

                    Archaeologia, London, Vol.XLIV, p.235,pp.242 et seqq.


McCaskill, M. (1996), Personal communication, The Flinders University of South Australia,  
                      Adelaide, an Expert Opinion. A copy is held by The Western Australian Maritime      
                      Museum under File MH 31/92.


McIntyre, K.G. (1977), The Secret Discovery of Australia, Portuguese Ventures 200 years  
                      before Captain Cook, Adelaide, London, pp.75-77, pp.362-365


Mills, J.V. (1930), Transl. And Notes, Eredia's Description of Malaca, Meridional India and
                      Cathay, M.S. 1613, Royal Asiatic Society Journal, Malayan Branch, Vol.VIII,

                      Pt. 1. (Singapore), Intro. p.1-6, Notes pp. 184-199.


Muller, K. (1993), Indonesian Spice Islands, the Moluccas, Singapore, p.32, pp.42-3, p.55


Napier, W. et al, (1973), Eastern Islands, Southern Seas, a History of Discovery and       
                       Exploration, London, p.80, pp.83-85, pp.132-3


Peters, N.H. (1992), The Iberian Charting of the North Coast of Australia in the

                      Sixteenth Century, unpublished thesis, pp.35-6, Ch.4 (Marginalia) pp.78-120.


Spate, O.H.K. (1957), Meanjin, Vol.16, No.2, Reprinted in Spate (1965),

                     Let Me Enjoy, A.N.U, Canberra, pp.249-266; p.258-60, p.263.


Ward, R. (1987), Finding Australia, Victoria, pp.84-5.


List of Plates

Plate 1


Map by Emanuel Godinho de Eredia, dated 1602

Plate 2       Map of Melville Island, Northern Territory, Australia

Plate 3       Part of the Eredia map (Plate 1) being the map of Ouro Island

Plate 4       Map of the northwest section of Melville Island.  Scale 1: 250,000

Plate 5       Part of the Eredia map (Plate 1) being a map of Luca.Antara Island

Plate 6       Map of Bathurst Island, Northern Territory, Australia

Figure 1.    Methods to compare antique maps against modern maps:

                   Triple Connections, Gridlock, triangles and trapezoids

I record my gratitude for the late Murray McCaskill's constant support and encouragement in this research, and likewise to Peter Grigg, LFRAIA, Architect, who served as a Sub. Lt. RANR during WW2 as a Watch Keeping Officer, in corvettes, with navigation duties.

Noel Peters

1/16 Cowan Road,

St. Ives. NSW 2075