thumb|Fragment of [[George Powell (sealer)|George Powell's 1822 chart of the South Shetland Islands showing the phantom Middle Island (bottom right) in Bransfield Strait, Antarctica]]
thumb|The [[Zeno map of 1558 showing Frisland – a phantom island in the North Atlantic]]
thumb|The phantom island of Kianida or Cianeis in the [[Black Sea on a fragment of the 1467 Nicolaus Germanus edition of Ptolemy's Geography]]
A phantom island is a purported island which has appeared on maps but was later found not to exist. They usually originate from the reports of early sailors exploring new regions, and are commonly the result of navigational errors, mistaken observations, unverified misinformation, or deliberate fabrication. Some have remained on maps for centuries before being "un-discovered".
Unlike lost lands, which are claimed (or known) to have once existed but to have been swallowed by the sea or otherwise destroyed, a phantom island is one that is claimed to exist contemporaneously, but later found not to have existed in the first place (or found not to be an island, as with the Island of California).
Examples
Some may have been purely mythical, such as the Isle of Demons near Newfoundland, which may have been based on local legends of a haunted island. The far-northern island of Thule was reported to exist by the 4th-century BC Greek explorer Pytheas, but information about its purported location was lost; explorers and geographers since have speculated that it was the Shetland Islands, Iceland, Scandinavia, or possibly nonexistent. The island of Hy-Brasil was sometimes depicted on maps west of Ireland, but all accounts of it have been fanciful.
Some phantom islands arose through the faulty positioning of actual islands, or other geographical errors. Pepys Island was a misidentification of the Falkland Islands. The Baja California Peninsula and the Banks Peninsula in New Zealand each appear as islands on some early maps, but were later discovered to be attached to their mainlands. Isle Phelipeaux, an apparent duplication of Isle Royale in Lake Superior, appeared on explorers' maps for many years, and even served as a landmark for the border between the United States and the territory that would become Canada, before subsequent exploration by surveyors determined that it did not exist.
Sandy Island appeared on maps of the Coral Sea beginning in the late 19th century. Purportedly, it existed between the Chesterfield Islands and Nereus Reef near New Caledonia; however, it was "undiscovered" in the 1970s. Nonetheless, it continued to be included in mapping data sets into the early 21st century, until its non-existence was re-confirmed in 2012.
Other phantom islands are misidentifications of breakers, icebergs, fog banks, pumice rafts from underwater volcanoes, or optical illusions. Observed in the Weddell Sea in 1823 but never again seen, New South Greenland may have been the result of a superior mirage. Some such as Thompson Island or Bermeja may have been actual islands subsequently destroyed by volcanic explosions, earthquakes, submarine landslides, or low-lying lands such as sand banks that are no longer above water. Pactolus Bank, visited by Sir Francis Drake in 1578, may fit into this former sand bank category.
In some cases, cartographers intentionally include invented geographic features in their maps, either for fraudulent purposes or to catch plagiarists.
List of phantom islands
{| class="sortable wikitable"
! width=150|Name
! width=150|Date of<br>alleged<br> discovery
!class=unsortable| Notes
|-
|Anaa-ti
|Unknown
|Location given as 22° 15′ S, 137° 30′ W, in the Tuamotus. Believed to be a mistaken sighting of a nearby island.
|-
| Buss Island || 1578 || Found in the waters near Greenland, in which Martin Frobisher, the leader of the island-finding expedition, probably made a mistake in dead reckoning and mistook optical effects near Greenland for a new island.
|-
| Byers's Island
|
|
|-
| || 1510 || A misconception about the Baja California Peninsula being an island due to an assumption that the Gulf of California was instead a strait separating California from the rest of the Americas.
|-
| Candyn
|
|
|-
| Cassiterides || || Ancient source of Phoenician tin. Exact location unknown but thought to have possibly referred to now silt-connected islands within the marshes of Brière.
|-
|Clark's Reef
|Unknown
|Discoverer unknown, location given as 8° 18′ S, 139° 50′ (or 52′) W. Admiral Du Petit Thouars could not find the reef, sounding 200 fathoms.
|-
| Isle of Demons || 1508 || Probably a relocated version of the island of Satanazes (see island below).
|-
|Denia/Davia
|Unknown
|Location given as 41° or 42° S, 20° E, in close vicinity of Marzeveen/Maarseveen. Not seen on modern maps.
|-
| Ernest Legouve Reef || 1902 || A reef supposedly found by the captain of the French ship, Ernest Legouvé, which is near the exact location of the fictional Lincoln Island, the main setting for Jules Verne's book The Mysterious Island, also appearing in In Search of the Castaways.
|-
| Estotiland || 1558 || An island appearing on the Zeno map at the current location of Labrador.
|-
|Faith Island
|Unknown
|Location given as 21° 10′ S, 138° 52′ W, in the Tuamotus. Believed to be a mistaken sighting of a nearby island.
|-
| Filippo Reef || 1886 || This reef, part of the Line Islands, was first seen by the ship Filippo and was seen again in 1926 when both ships saw breakers in the same area, suggesting a depth of . Current observations show the reported location to have a depth of , and the nearest shallow seamount is about deep, disproving the existence of the island.
|-
| Fonseca Island || 1544 || An island sighted east of Barbados.
|-
| Frisland || 1558 || Another island on the Zeno map, possibly a renamed Iceland.
|-
| Ganges Island || || A nonexistent island off the coast of Japan to the southwest of the Shatsky Rise.
|-
|Isle Grande
|1675
|Discovered by Antonio de la Roche. Roche only passed the island on its eastern side. Various locations given, all at 45° 15′ S, but otherwise differing at 38° 15′ W (per la Roche), 45° 30′ W, and 35° 30′ W; considered uncertain by 1808. Possibly a mistaken sighting of a projecting headland from South America, as la Roche never saw the other side.
|-
|Kettendyk's Droogte
|Unknown
|Location given as 33° S, 4° 25′ E, northeast of Tristan da Cunha and west of South Africa. Unsuccessfully searched for, and not seen on modern maps. According to Bulgarian geomorphologist Dinyo Kanev, probably destroyed by sea in the Middle Ages.
|-
|Krusenstern Rock
|1804
|Reported as a breaker at 22° 15' N, 175° 37' W. Capt. R. Suffern of the Craigerne reported that he was at these exact coordinates in 1897 but there was no sign of the rock.
|-
|L'Enfants Perdu Islands
|1768
|Discovered by Bougainville, variously placed at 14° 16′ S, 177° 23′ W or 14° 20′ S, 176° 40′ W. Found doubtful in 1875 after searches found no land in the area. Possibly a mistaken sighting of the Horne Islands.
|-
|Minnehaha Rock
|1879
|Sighted by Capt. Beckwith of the Victoria at 25° 50' S, 106° 20' W. No subsequent sightings have been made.
|-
|Nachtegal Rock
|1861
|Seen by at 40° 20′ S, 52° 55′ E. Last seen in charts in 1878. of the United States Department of Defense.
|-
| Terra Nova Islands || 1961 || Thought to lie off Oates Coast, East Antarctica.
|-
| Thompson Island || 1825 || An island in the south Atlantic Ocean discovered by the whaling ship captain George Norris; it has not been seen since 1893.
|-
| Thule || || A mythical island in the far north, possibly at or above the Arctic Circle, mentioned in many works from the Roman and Medieval period. Sources in antiquity placed Thule several days travel north of Great Britain visible from Orkney; or north of Scythia. More modern scholars have suggested Thule may have been Ireland; the Estonian island of Saaremaa; or the Norwegian island of Smøla.
|-
|Tiburones
|Unknown
|Captain D'Urville asked the residents of Uapoa about this island in August 1838. They claimed it existed, reporting it had high land, one sandy beach which could be approached in good weather, and a single male inhabitant (the others having fled). Location given as 11° S, 143° W; noted doubtful in 1851 and not seen on modern maps.