How the “Phantom” Island That Fooled Maps for 100 Years Was Exposed

A map can look authoritative while still carrying an old mistake. For more than a century, an elongated landmass called Sandy Island appeared in charts and digital geographic databases northwest of New Caledonia. It had a name, a recognizable outline and even an apparent elevation in some scientific models. There was only one problem: no island occupied the location. The disappearance of this phantom feature was not a tale of a landmass sinking beneath the waves, but a revealing lesson in how uncertain observations can survive through generations of copied and interconnected data.

A Phantom Island in the Coral Sea

Sandy Island was commonly plotted near 19 degrees south and 160 degrees east, approximately 500 kilometres northwest of New Caledonia. It lay between the Chesterfield Islands and Nereus Reef in the eastern Coral Sea—at least according to numerous maps, coastline databases and seafloor models.¹

The feature was sometimes called Île de Sable, the French equivalent of Sandy Island. On modern digital maps it appeared as a long, thin shape, while Google Earth once displayed a dark polygon in the same area. To a casual observer, these repeated appearances could seem like independent confirmation. In reality, many of the products were inheriting information from the same underlying coastline source.

Sandy Island belongs to a broader category known as phantom islands: landforms that were reported, charted and sometimes widely accepted before later investigation showed that they did not exist at the recorded position. Such errors were especially understandable during eras when navigators estimated longitude imperfectly, observations were made in poor weather and charts had to include possible hazards before they could be conclusively surveyed.

How the Error Entered the Historical Record

Older maps contain several features bearing names such as “Sandy Island,” but researchers caution that many were smaller and located elsewhere. The specific phantom island involved in the 2012 controversy is most clearly connected to a reported sighting by the whaling ship Velocity in 1876. A British Admiralty chart published in 1908 showed an island attributed to that report, with a shape and position similar to those later reproduced in digital data sets.¹

Nineteenth-century navigators often had to report anything that might endanger a vessel. Breakers, discoloured water, floating material or a poorly observed reef could all be recorded as potential hazards. In an age before continuous satellite observation and comprehensive sonar surveys, retaining a doubtful feature could seem safer than deleting a real obstruction.

Later hydrographic charts placed the notation ED beside Sandy Island. The abbreviation meant “existence doubtful,” signaling that repeated efforts had failed to verify the feature. This was not a declaration that the island certainly existed; it was a warning that the original report remained unresolved.

1. 1876

   The reported sighting

   The whaling ship Velocity was credited with observing the feature that later became the familiar Sandy Island outline.

2. 1908

   Appearance on an Admiralty chart

   A British chart displayed an island whose location and shape closely resembled the feature later copied into global digital databases.

3. 1974

   Removed from French charts

   The French Hydrographic Service deleted Sandy Island from its official charts after an aerial reconnaissance failed to support its existence.

4. 1985

   Removed by Australian hydrographers

   The Australian Hydrographic Service also treated it as a phantom after surveys showed no island or shallow reef at the recorded location.

5. October 2012

   Confirmed by direct measurement

   The Southern Surveyor crossed near the supposed island and recorded deep water rather than exposed land.

6. 2013 and afterward

   Digital databases corrected

   Scientific coastline, terrain and mapping products were updated to remove the inherited feature.

Why Sandy Island Survived After Experts Rejected It

The most important part of the mystery is not how an uncertain nineteenth-century sighting entered a chart. It is how the feature persisted in modern databases after French and Australian hydrographic authorities had already removed it from navigational charts.

A major link in that chain was the World Vector Shoreline database, commonly abbreviated WVS. It was created by digitizing coastlines from paper charts so that researchers and mapping software could use a consistent global shoreline. This was enormously useful, but digitization also preserved some mistakes contained in the source material.¹

The World Vector Shoreline data influenced later products, including global elevation and coastline compilations. GEBCO—the General Bathymetric Chart of the Oceans—reported that its own terrain model inherited Sandy Island indirectly through data sets that relied on WVS.²

This explains why the island appeared to have support from several scientific sources. Some sea-surface-temperature and bathymetric products displayed a blank area or land value at its location because a pre-existing land mask told the software that the pixels represented land. Those products were not independently detecting an island. They were processing their measurements according to a coastline boundary already assumed to be correct.

The 2012 “Undiscovery”

In October 2012, scientists aboard the Australian research vessel Southern Surveyor were conducting a 25-day expedition to study the tectonic history of the eastern Coral Sea. Sandy Island was not the expedition’s original destination. The discrepancy became important when researchers noticed that the island appeared in global scientific data and Google Earth but was absent from the legal hydrographic charts used by the ship’s master.¹

The vessel passed approximately five kilometres south of the mapped island, following the nearby bathymetric high shown on official charts. The crew saw open ocean. More decisively, the ship’s multibeam sonar recorded a minimum depth of about 1,300 metres along its track. Earlier Australian soundings in the surrounding area had also measured depths ranging from approximately 1,488 to 2,353 metres.

Those measurements ruled out not only an exposed island but also the idea of a shallow reef lying just beneath the surface. Satellite imagery offered the same broad conclusion: there was no convincing colour change or biological signature resembling nearby shallow reefs. The location was deep ocean, not a recently submerged tropical island.

The scientists jokingly described their finding as an “undiscovery.” The phrase captured public attention because exploration is normally associated with adding features to maps. Here, fieldwork improved knowledge by removing one.

Did Sandy Island Sink?

No evidence indicates that a large island existed there in recent history and later sank. A landmass approximately 25 kilometres long could not reasonably disappear beneath more than a kilometre of water over such a short geological interval without leaving substantial evidence.

Researchers also noted that the area lacks the active volcanic setting expected if a young volcanic island had rapidly emerged and then subsided. The deep seafloor measurements and satellite observations are therefore inconsistent with stories of an island being eroded, swallowed by the sea or destroyed by a sudden catastrophe.

What Might the Crew of the Velocity Have Seen?

The original cause cannot be reconstructed with certainty. A navigational mistake, an incorrectly copied coordinate or confusion with another reef may be enough to explain the record. One especially interesting proposal is that the sailors encountered a pumice raft.

Pumice forms when gas-rich volcanic material cools into highly porous rock. Some pieces are buoyant enough to float, and large eruptions can cover vast areas of ocean with pale, tightly packed fragments. From the deck of a nineteenth-century vessel, a distant pumice raft—particularly one mixed with waves, foam or floating debris—could resemble a low sandy shore.

Ocean-current studies have shown that pumice released near Tonga can travel westward through the region between Fiji and New Caledonia on its way toward Australia. One modeled pumice route passed within about 20 kilometres of Sandy Island’s supposed position. Researchers therefore consider a raft plausible, but it remains a hypothesis rather than a proven explanation.¹

Why Satellites Did Not Automatically Fix the Map

It is tempting to assume that satellites photograph every part of Earth and translate those images directly into a flawless map. Modern mapping is more complicated. A digital map may combine optical imagery, radar observations, elevation models, depth soundings, gravity measurements, historical shorelines and manually edited labels.

Satellite-derived gravity data can help scientists estimate the broad shape of the seafloor because underwater mountains subtly affect Earth’s gravitational field and the height of the ocean surface. However, this technique cannot by itself determine whether a feature rises above sea level. Near the supposed Sandy Island location, gravity data suggested elevated seafloor nearby, but not an exposed island.²

Google Earth’s former dark polygon was also not a photograph of mysterious land. The coastline data instructed the system to treat the area as land, while no normal land imagery was available to fill it. The visual gap was therefore a symptom of conflicting data—not proof that something had been concealed.

What Sandy Island Teaches About Digital Information

Sandy Island is often treated as a charming geographical curiosity, but its deeper lesson applies to nearly every field that relies on large data collections. Digital information can be precise without being correct. A coordinate may contain many decimal places, a model may use advanced mathematics and a feature may appear in polished software, yet the result can still depend on an unverified historical input.

• Repeated information is not always independent evidence. Several websites or data sets may be copying the same original source.
• Uncertainty should travel with the data. A chart marking a feature as doubtful communicates more honestly than a database that converts it into definite land.
• Field observations remain essential. Satellite models are powerful, but direct depth measurements were crucial in resolving this case.
• Corrections must move through entire data chains. Fixing one database is not enough when derived products continue to preserve an older version.
• Maps are models, not the territory itself. Every map simplifies reality and reflects the sources, methods and decisions of its makers.

Are There Other Phantom Islands?

Sandy Island is not unique. Maritime history contains numerous islands, reefs and coastlines reported under uncertain conditions and later removed from charts. Some resulted from mistaken coordinates, clouds on the horizon, mirages, floating ice, pumice, duplicated discoveries or confusion between similarly named places.

Modern surveying has resolved many such errors, but mapping the oceans remains an enormous undertaking. Satellites can observe the sea surface globally, yet detailed measurements of the seafloor still depend heavily on ships equipped with sonar. Large areas have not been mapped with the same resolution routinely available on land.

The meaningful question is therefore not whether modern cartographers are careless. It is how mapping organizations identify uncertainty, record the origins of their data and correct inherited errors without discarding useful information.

Frequently Asked Questions