Three places have been dug for the colonists: Roanoke Island since 1895, the Chowan shore of Bertie County (the mainland at the head of Albemarle Sound) since 2012, and Cape Creek on Hatteras (the site of the Croatoan town) since 1993. A fourth, the Chesapeake, has been dug for other reasons. The published site reports and technical studies (the 1962 report of J. C. Harrington, the National Park Service archaeologist who dug on Roanoke Island; the British Museum's 2012 bulletin; the 2019 article by Charles Ewen, an archaeologist at East Carolina University, and Farrell; Mary Ellen Hodges's 1998 report on the Chesapeake site; the First Colony Foundation's own PDFs and web statements; the National Park Service's dig summaries) and every press report of the unpublished work through 14 September 2026 are read here, and each claim is checked against the site's own words. Where a claim rests on an interview, the text says so.
Seventy-nine years of excavation at Fort Raleigh National Historic Site, the Park Service's ground at the north end of Roanoke Island (Harrington 1947 to 1953 and 1965; the Park Service in the 1980s; Ivor Noël Hume, the Colonial Williamsburg archaeologist, 1991 to 1995; the First Colony Foundation, a nonprofit group formed in 2004 to search for the Roanoke colonies, 2006 to 2021) have produced a small earthwork, a metallurgical workshop beside it with the debris of the ore assays (tests of ore for metal) made by Lane's colony, and Native cache pits with 1580s trade copper. Harrington's conclusion of 1962 stands: "the settlement site has not been located". No house, palisade line, burial or midden (refuse heap) of the 1587 colony has been identified anywhere on the island. The First Colony Foundation's reinterpretation of the earthwork (2022, revised 2024) assigns it to Lane's colony on the strength of assay debris under its rampart; the palisade "very Fortlike" that White found in 1590 has not been seen in the ground. The only feature ever proposed as later than 1586 is a group of three fire pits in a watchtower footprint, which the Park Service reported in 2021 with "no way to know" whether they are 1585 or 1587. The north end of the island "has eroded extensively since the sixteenth century", in the words of the marine archaeologist Gordon Watts (2008, in the Foundation's 2007 report), and the settlement may lie under the sound. The absence of the settlement site therefore bears neither for nor against any reading of the colonists' fate.
In 2012 the British Museum examined two paper patches on White's manuscript map of the sounds, drawn in 1585 to 1586 and now in the museum (1906,0509.1.3), at the request of a First Colony Foundation board member. The technical study (Ambers, Russell, Saunders and Sloan, British Museum Technical Research Bulletin 6, pp. 47 to 54) found that the second patch, at the head of Albemarle Sound where the Chowan and Roanoke meet, conceals "a boldly coloured symbol apparently indicating a fort or signifying the intended location of one", and that a fainter symbol was redrawn on top of the patch. The authors decline to interpret it: "the detailed interpretation of this symbol is beyond the scope of this study", and "it is not clear if this shape is intended as a symbolic representation or as an actual depiction of the ground plan of a fort or of an intended, but unbuilt, fort". The map is a document of 1585 to 1586. It records that someone in Lane's colony marked, and then hid, a fort position at the Chowan mouth. That is consistent with Lane's plan to move by way of "the Riuer of Choannak" (section 1) and with the colonists' "50 miles into the maine". It is not evidence that the 1587 colonists went there. Horn's "conclusive proof" (2012) is an inference from the map to the colony.
The Foundation dug Site X, beside the Native town of Mettaquem that White's map marks at the mouth of Salmon Creek, from 2012 to 2017: 53 two-metre units, in a plough zone (topsoil that centuries of ploughing have turned over, mixing whatever it holds) mixed with some 275 pounds of Native pottery, produced "40 sherds representing perhaps 6 or 7 vessels of 16th century Surrey-Hampshire Border ware; 8 sherds representing a North Devon plain baluster jar" (a sherd is a fragment of a pot; Border ware is an English kitchen pottery made on the border of Surrey and Hampshire), two tenter hooks (for stretching cloth), an aiglet (the metal tip of a lace), priming pans from gun locks, a buckle and, announced in 2026, a lead seal of the kind fixed to bolts of cloth at Augsburg, of the late sixteenth century. Site Y (the state's site number 31BR49), a short distance away, was tested from December 2019 in 72 five-foot squares and produced a more varied English assemblage (Border ware, North Devon baluster jar, London and Essex redwares, Spanish olive jar, German (Frechen) and French (Martincamp) stoneware) with no Native material; no sherd count has been published. Neither site has produced a structure, a feature, a burial or a sealed sixteenth-century deposit. The dating rests on typology, the dating of pottery by its style: Border ware "declined precipitously after about 1630", and the glass beads and clay tobacco pipes that mark later trade are absent. In its 2023 volume the Foundation put the argument in numbers. Site X, with 2,282 square feet excavated against 10,000 to 30,000 at a sample of Virginia sites, "has five to ten times as much Border ware as the early Virginia sites, excepting the first years of settlement at Jamestown"; Eden House, the governor's seat built on the Chowan no earlier than about 1663, yielded "just four sherds from a single vessel of a late type of Border ware"; and the absence at Site Y of "colonial tobacco pipe stems with large bore holes" is offered against any later seventeenth-century occupation. An answer to the Foundation now has to meet that comparison rather than the bare presence of the ware. The same chapter states the limit of the claim in the words of Nicholas Luccketti, the Foundation's archaeological director: "no artifacts undeniably of sixteenth-century European origin. No features from English buildings, no burials. We did not find a fort. Nor did we claim that we had found the site of the main group's relocation". It canvasses the colony's pinnace "moored nearby" as one explanation of the site, and its sonar survey of Salmon Creek recorded several small submerged craft and one larger target near the creek's mouth, none of them examined. Critics (Ewen, Henry Wright of the University of Michigan, Horton, Noël Hume) answer that a mixed plough-zone assemblage cannot be dated to within a generation, that a Jamestown-era origin between 1607 and 1650 is not excluded, and that Lane's 1586 voyage up the Chowan is not excluded either. No peer-reviewed site report exists for either site; the fullest account is a popular volume edited by Eric Klingelhofer, one of the Foundation's archaeologists (Excavating the Lost Colony Mystery, 2023), and the Foundation's own dig-reports page lists nothing for Bertie County.
The Foundation's statements have grown more confident between 2017 and 2026 without any new data being published. "We do not have the smoking gun" (2017, repeated 2023); "Site Y does not solve the Lost Colony mystery" (December 2020); "we are 100 percent sure that we have two sites that are related to the Roanoke Colonists" (Luccketti, December 2020); the colonists are "still lost" and "more work is needed to confirm it" (June 2024); and on 4 June 2026 a web statement headed "The Lost Colony is Found! How We Did It." which opens "This is not another hoax or false claim" and concludes that Site X and Site Y "were part of a larger, dispersed settlement of farms along the Chowan River, where 117 immigrants established their New World home". The statement reports no new fieldwork, no new dates and no new counts; the Foundation's news index carries nothing between 17 October 2024 and 4 June 2026. What the two sites show is a small, typologically coherent group of late sixteenth or early seventeenth-century English kitchen and storage ceramics with a few dress and gun fittings, at two spots on the Chowan shore beside a Native town, consistent with a small English presence there for a season or more, of a date between 1586 and about 1650. That is a finding in its own right, but it is not evidence of a colony of 117 people.
David Phelps, an archaeologist at East Carolina University, excavated the Croatoan town site at Cape Creek, near the village of Buxton, from 1993 to 1998, dated it 1650 to 1720, found a signet ring he took for gold and a snaphaunce lock (an early form of flintlock), and never published. The ring, re-examined in 2017 by X-ray fluorescence, a test of a metal's composition, is brass, and it lay in a context of 1670 to 1720 (Ewen and Farrell, North Carolina Historical Review 96 (4), 2019, pp. 408 to 425). Mark Horton, a British archaeologist, with the Croatoan Archaeological Society, a local group on Hatteras founded by Scott Dawson, dug from 2009 and found a rapier hilt, a slate writing tablet, a Nuremberg counter (a brass token used in reckoning accounts), gun parts, aiglets and coin weights dated 1644 and 1648, most of it in seventeenth-century contexts; Horton himself said in 2020 that the objects could have arrived by trade, and Ewen reads the stratigraphy (the order of the layers) as seventeenth-century.
The claim that has most altered the public account since 2024 is the hammerscale. In an excavation of summer 2024 Horton's team found the flakes that fall from iron being forged, and in January 2025 the first report spoke of "buckets" of it. The sequence since then is: a Daily Mail article of 11 May 2025 ("SOLVED", Horton: "Once and for all, this smoking gun evidence"); Live Science of 10 June 2025, where Horton describes "hammer scale in a pit underneath a thick shell midden that contains virtually no European material in it at all" and the writer notes that the finding "has not yet been published in a peer-reviewed journal"; a talk at the Coastal Studies Institute on Roanoke Island on 26 March 2026 titled "The Smoking Gun?", after which Horton told the Outer Banks Voice that "archaeologists have difficulty dating material to differentiate 50 years on either side" and that the objects "could be 17th-century trades from Jamestown"; and a WAVY report of 1 June 2026 in which Scott Dawson of the Croatoan society says that four radiocarbon results have come back, "All of them were from the 1580s", among them a date on the teeth of a deer whose jaw held a musket ball, which Dawson himself allows may have come from Lane's men of 1585. The argument is sound in principle: hammerscale is waste that was not traded, so iron was forged on Croatoan, and forging was a European craft. Whether the layer is of the 1580s, the 1600s or the 1650s is the whole question, and it turns on stratigraphy and dates that have not been published in any form a reader can check. As of 14 September 2026 no journal article, monograph or technical report by Horton or the society on the hammerscale, the radiocarbon dates or the 2009 to 2024 excavations exists. Ewen has said he wants to see a hearth; Whitney Leeson, an archaeologist at Roanoke College, wants slag and charcoal. Until the section drawings and the dates are in print, the hammerscale cannot be counted as a dated result.
Quinn's thesis that the main body went to the Chesepian towns on the Lynnhaven and Elizabeth rivers has never been tested by excavation, and the one large Chesepian site that has been fully reported, Great Neck (site 44VB7, in what is now Virginia Beach), yielded no European artefact earlier than the nineteenth century and a late Native occupation that the excavator did not date after sustained European contact (Hodges 1998). Strachey says the district was re-peopled by Powhatan's own people by 1611 (p. 105). No excavation supports the thesis, and none has been designed to test it.
In 2023 to 2024 the Foundation found the village of Roanoac, an Algonquian-speaking community like the other peoples of the sounds, in the Elizabethan Gardens, a garden beside Fort Raleigh on Roanoke Island, with a drawn-copper ring of European metal; it is a contact site of 1584 to 1586, abandoned when Lane left. The Lumbee descent claim has no archaeological component. The Lost Colony DNA Project, founded in 2007 to find colonists' descendants by surname lineages, is listed as inactive and reported no match.
Any of the following would move the question decisively; none has been found since excavation began in 1895:
The archaeology as it stands gives two places with a small English footprint of the right period, one on the banks and one on the mainland at the west end of the announced fifty miles, and nothing at all for the Chesapeake or for a massacre site. That is also the distribution the documents in sections 1 to 3 describe.