Roanoke, August 1590: the colony, the carving, and the empty fort

On the afternoon of 18 August 1590, John White and a landing party from two pinnaces of an English privateering fleet walked up the north end of Roanoke Island toward the place where he had left his colony three years and three months earlier. The settlement he had governed in 1587 had been a cluster of houses inside the ruined earthworks of the older Lane garrison. What he met that afternoon was something else. The houses had been “taken downe.” In their place stood a high palisade of great trees, with curtains and flankers, “very Fort-like.” On a tree at the brow of the bluff above the sound shore, “curiously carved,” were three Roman letters: CRO. On a chief post on the right side of the entrance to the new palisade, “in fayre Capitall letters,” was a single word: CROATOAN. There was no cross. Before he had sailed for England in 1587 he had agreed a secret token with the colonists. If they moved, they would carve the name of the place they had gone. If they moved under duress, they would add a cross. The post carried the name. It did not carry the cross.

White wrote it down himself, in a narrative Richard Hakluyt published in 1600. What the carving means, what happened in the three-year gap between his August 1587 departure and his August 1590 return, and what finally became of the 115 to 118 men, women, and children he had left on the island, is what the 438 years since 1590 have not settled.

The documented account

Sir Walter Raleigh’s 1584 patent from Queen Elizabeth I authorised him to colonise in her name. Amadas and Barlowe’s 1584 reconnaissance returned from the Outer Banks with two Algonquian men: Manteo of the Croatoan people of present-day Hatteras Island, and Wanchese of the Secotan on the mainland. Sir Richard Grenville’s 1585 expedition left a 108-man military garrison under Governor Ralph Lane on Roanoke Island. Lane’s garrison killed the Secotan weroance Wingina (Pemisapan) in June 1586 in a collapsed alliance, then evacuated with Drake’s passing fleet. A small holding party of fifteen to eighteen men, left on Grenville’s instructions, vanished before the 1587 expedition arrived.

The 1587 expedition was a civilian colony, chartered on 7 January as the “Cittie of Raleigh in Virginea” with White as Governor and twelve named Assistants. The intended landing place was the Chesapeake Bay; Roanoke was to be a stop on the way, to collect Grenville’s holding party.

The fleet sailed from Plymouth on or about 8 May 1587 under the Portuguese pilot Simon Fernandes. White’s manifest lists about 117 colonists, including seventeen women and eleven children; secondary counts range from 115 to 118. The fleet stopped at Croatoan Island on or about 22 July, where Manteo, returning home a second time, re-established contact with his people. On 25 July the ships reached Roanoke.

Then the first decisive moment of the case. White intended to take the colonists onward to the Chesapeake. Fernandes refused, citing the season, and “would land all the Planters in no other place.” The colonists came ashore on Roanoke, not at the bay they had been chartered for.

George Howe was killed by Secotan warriors while crabbing alone, the only colonist death on record before the disappearance. On 13 August 1587, Manteo was baptised at Raleigh’s instruction and titled “Lord of Roanoke and Dasamongueponke.” On 18 August 1587, White’s daughter Eleanor Dare gave birth to Virginia Dare, the first English child born in the Americas. A second English child, born to the Harvie family, followed within days.

The supply position was already thin. The colonists pressed White to sail back to England and plead the colony’s case. He sailed with Fernandes on 27 August 1587, ten days after his granddaughter’s birth, and reached England by 5 November. Before he left, they agreed on the secret token, recorded in his 1590 narrative as “the secret token agreed upon betweene them & me at my last departure from them.” Everything in the case turns on whether the cross was added.

The relief did not come in 1588. A small April voyage was attacked by French privateers and forced back; the larger relief was prevented by the Spanish Armada emergency. White’s second attempt in spring 1590, aboard a privateering expedition that prioritised prize-taking in the Caribbean, did not sight the Outer Banks until mid-August. Three years and three months elapsed between his departure and his return.

The pinnaces reached the Outer Banks on 15 August 1590. A surf accident in which Captain Spicer drowned with six others delayed the landing. White reached the abandoned settlement on 18 August, his granddaughter Virginia’s third birthday. He found “the houses taken downe” and “a high palisado of great trees, with cortynes and flankers very Fort-like,” and “in fayre Capitall letters was graven CROATOAN without any crosse or signe of distresse.” His buried property had been dug up; his books were “torne from the covers” and his armour “almost eaten through with rust.” His own reading: “I greatly joyed that I had safely found a certaine token of their safe being at Croatoan.”

He never reached Croatoan Island. In the night a cable parted, three anchors were lost, and the master of the larger pinnace overruled him. They turned for the Caribbean and then for England. White reached home by 24 October 1590. He never sailed for Virginia again. He died around 1593, in obscurity. He never found the colonists.

After 1590 the official record thins. Raleigh fell from royal favour on the accession of James I in 1603 and his patent lapsed. The Jamestown expedition of 1607 was instructed to inquire after the 1587 company. Two thin Jamestown-era strands survive at one or two removes from any English eyewitness, both treated in the hypotheses below: John Smith’s Generall Historie (1624) and a Strachey manuscript completed around 1612 and first printed in 1849.

The evidence

The case rests on one first-person English narrative, one physical clue inside it, one period English map with two later patches, a 1612 Jamestown-era manuscript, and four strands of modern archaeology. None has produced a positive identification of the 1587 company.

The carved inscriptions on the post and the tree. The post read CROATOAN; the tree read CRO; neither carried the duress cross. The houses had been deliberately dismantled, not burned, the wood probably reused for the new palisade. The carvings establish that someone left the settlement deliberately, pointed at Croatoan, and followed the non-distress protocol. They do not establish when they were carved, whether the palisade was raised by the colonists or a later occupant, or whether Croatoan was the terminus or a way-stop.

John White’s 1590 narrative. The only first-person source for the landing and the return. White was not present during the three-year gap.

The “La Virginea Pars” map and the 2012 lightbox examination. White’s 1585 working coastal map, held by the British Museum (1906,0509.1.3), carries two later paper patches. A 2012 non-contact lightbox examination by the museum, initiated by Brent Lane of the UNC Kenan Institute, found the northern patch pasted over what appears to be a Renaissance-style fort symbol in red and blue inks at the junction of the Roanoke and Chowan rivers in present-day Bertie County, about fifty miles inland. The symbol’s purpose is not established. It may indicate a planned fort never built, a built fort concealed for reasons of patent claim or Spanish counter-intelligence, or a relocation destination.

Site X and Site Y in Bertie County. The First Colony Foundation, on the strength of the hidden symbol, has excavated since 2012 at Site X (Salmon Creek and Albemarle Sound) and more recently at Site Y on the Chowan River. The combined assemblages include Surrey-Hampshire Border Ware, North Devon baluster jar, London redware, Spanish olive jar, two iron tenter hooks, and an Elizabethan-style aiglet, consistent with a small late-16th-century English presence. Excavation director Nicholas Luccketti has said on the record: “We do not have the smoking gun… Nor do we claim to have found the main relocation site.” Pottery dates span several decades. No burial, no signed object, no identifiable colonist artifact has been recovered.

Cape Creek and the Hatteras Island excavations. Systematic work on Hatteras Island, where the 16th-century Croatoan capital stood, has run since 1995 under David Phelps, then Mark Horton, and the Croatoan Archaeological Society under Scott Dawson. Finds include a swept-hilt rapier fragment, a gunlock, a brass apostle or gunpowder flask, a lead pencil and slate, a bronze signet ring (Phelps, 1998), Border Ware and North Devon ceramics, and square postholes consistent with European-style construction within the village pattern. A 2025 Horton-Dawson report, dated via animal-tooth radiocarbon at UC Irvine, identified hammerscale in a stratigraphic layer dated to the late sixteenth century. The signet ring’s original Kendall attribution has been re-evaluated and may be 17th-century. The 2025 report is not yet peer-reviewed. The Hatteras assemblage is the richest late-Elizabethan English material in any Algonquian village context yet excavated. No 16th-century English burial has been documented on the island.

The Strachey passage. The 1849 first printing of the ca. 1612 manuscript made public the “slaughter at Roanoke” report and the Ritanoe survivors. The chain is Machumps to Strachey to manuscript to 1849 printing. Strachey is the closest the documentary record comes to a contemporary report of the colonists’ fate, and he is a quarter-century removed and writing on hearsay.

The Dare Stones, a documented twentieth-century hoax. Between 1937 and 1940, forty-eight inscribed stones purporting to record the colonists’ fate, including stones signed “EWD” for Eleanor White Dare, were presented to Haywood Pearce Jr. of Brenau College, Georgia. Forty-seven were brought in by William “Bill” Eberhardt, a Georgia stonecutter with a prior history of selling counterfeit Native American artifacts. Boyden Sparkes’s April 1941 Saturday Evening Post investigation established the closed circle of “discoverers” and linguistic anachronisms inconsistent with 1590s English. The forty-seven Eberhardt stones are definitively forgeries; the 1937 Hammond first stone alone has not been fully disproved. The Dare Stones are noted because the record includes them. They are not evidence of what happened to the colony.

Documented absences. No 16th-century English burial anywhere in the search area. No individually identifiable colonist artifact. No Spanish archival record naming Roanoke captives, despite repeated searches of the Archivo General de Indias in Seville.

Hypotheses and open questions

Each of the following is layer-3 speculation. None is asserted as fact.

Hypothesis A. Peaceful relocation to Croatoan Island (James Horn, A Kingdom Strange, 2010). Horn argues the record is consistent with the colonists, or some of them, having moved to Croatoan and integrated with Manteo’s people. The reading rests on the carving, the absence of the duress cross, Manteo’s diplomatic role, and the late-Elizabethan assemblage at Cape Creek. The 2025 hammerscale finding, if peer-reviewed, would strengthen it. Constraints: the Cape Creek material does not unambiguously date to 1587; no 16th-century English burial has been recovered; Croatoan may have been a way-stop.

Hypothesis B. Inland relocation toward the Roanoke-Chowan junction (First Colony Foundation, Site X and Site Y). The Foundation argues the colony, or some part of it, relocated to the inland location indicated by the hidden symbol. White’s narrative records the colonists’ pre-existing intention to “remove from Roanoak 50 miles into the maine.” Constraints: Luccketti says they do not have the smoking gun; pottery dates are diffuse; no fortification, burial, or identifiable colonist material has been recovered.

Hypothesis C. The Strachey “slaughter at Roanoke,” a Powhatan attack ca. 1607 (Horn 2010, on the 1612 manuscript). Strachey reports Machumps’s testimony of twenty years of peaceful inland residence, then attack by Wahunsenacah’s warriors at his priests’ urging, with seven survivors preserved by Eyanoco at Ritanoe. Smith’s Generall Historie (1624) carries separate Powhatan testimony of “men clothed like me” inland. Constraints: chain of informant to secretary to manuscript to printing; Strachey never saw the events; no Chesapeake-region late-Elizabethan English settlement has been recovered.

Hypothesis D. Capture or destruction by Spanish forces. Spanish authorities sought to destroy English colonies in the 1580s; Vicente González’s 1588 reconnaissance reached the Outer Banks and reportedly sighted signs of English presence. Constraints: no Spanish source naming Roanoke captives has been found in the Archivo General de Indias; the 1590 condition of the settlement, dismantled with a non-distress carving, is inconsistent with a forced removal.

Hypothesis E. Epidemic, starvation, or weather-driven collapse. Lane’s 1585 garrison was near starvation within a year on the same island; the 1587 company was civilian; Old World epidemics were moving through Algonquian populations. Constraint: a collapsing company would not have had the labor to take houses down and raise a palisade of great trees. CROATOAN is the act of an organised, departing group.

Hypothesis F. The Lumbee descent claim. Hamilton McMillan’s 1888 argument that the Lumbee people of Robeson County, North Carolina, descend in part from the 1587 colonists via intermarriage with regional Algonquian peoples has been periodically revived. The Lumbee community itself describes a complex multi-tribal heritage and has not endorsed it. Modern scholars treat the narrative as not borne out by the record. Reported as historiography; not endorsed.

What remains unknown

The Roanoke company’s fate has not been positively established in 438 years.

What the record does not establish is everything in the gap between 27 August 1587 and 18 August 1590. The carving is the colonists’ own pointer to a place; it is not a sentence about the journey. None of the archaeology, the manuscripts, or the Spanish searches has produced a 16th-century English burial or an identifiable colonist artifact tying a name in White’s manifest to a site on the ground. The Algonquian peoples in the period record, Manteo’s Croatoan and the Secotan and Chowanoc, are documented historical communities whose descendants carry their own histories that do not reduce to the English question.

The strongest sentence the record supports is White’s own. In August 1590 he found the houses taken down, a high palisade of great trees, and the word CROATOAN in fair capital letters on a post, without any cross or sign of distress. That sentence has been in print since Hakluyt published it in 1600. Where the colonists actually went, and what became of them, is what the next 438 years have not settled. The case stays open on the carving and on the empty fort.

Sources

Primary

John Smith, Generall Historie of Virginia (London, 1624), cited through modern editions.

Secondary

James Horn, A Kingdom Strange (Basic Books, 2010); David Beers Quinn, Set Fair for Roanoke (UNC Press, 1985); Boyden Sparkes, “Writ on Rocke,” Saturday Evening Post (26 April 1941).