Last Updated: October 2026
Most castle facts are settled. A few are not. Construction dates drift between sources, a founding year is given three different ways, an operator’s own website contradicts itself, and a measurement turns out to describe something other than what everyone assumed. How a reference handles those cases says more about it than the facts everyone already agrees on.
This page collects real examples from across StoneKeep Atlas where the sources genuinely disagree, and shows what we did about each one. The rule is the same every time and it is set out in full in our editorial standards: when two credible sources conflict, we do not quietly pick the more convenient number, and we do not split the difference. We say the record is divided, we attribute each claim, and where a specific cannot be pinned down honestly we leave it out rather than invent certainty. Sieges carry their own disagreements, and our ledger of how castles fell names both sources wherever two of them conflict over how a siege ended.
Dates the record does not agree on
Château de Langeais: when was the keep built?
Langeais has one of the oldest stone keeps in France, and its date is contested. The traditional attribution, repeated by the Bibliothèque nationale de France, gives “around 994” and credits Fulk Nerra, Count of Anjou. Dendrochronology carried out on the near-identical keep at nearby Loches, however, points to the early eleventh century, roughly 1012 to 1035. Both are defensible, and they cannot both be exact. We attribute the keep to Fulk Nerra and date it qualitatively, “around the year one thousand,” and we do not print a single hard year in either direction.
Schloss Drachenburg: 1881 or 1882?
The foundation stone of this Rhineland fantasy palace is dated to 1882 by the operator’s own history and by the German-language record, while a competing 1881 start date circulates in secondary sources. That building span, 1882 to 1884, is well supported. Because the exact foundation year is genuinely divided, any article that mentions Drachenburg in passing avoids the single year and refers to the early 1880s.
Château d’Ussé: when did the d’Espinay family acquire it?
One widely repeated source gives 1455; the operator’s own press kit gives 1485 (“1485 Ussé passe à la famille d’Espinay”). A 1462 marriage into the Bueil family sits awkwardly between the two and adds to the tension. Rather than choose, we date the acquisition to “the 1480s” and keep the marriage chronology as the reason the earlier date is doubtful.
Mespelbrunn: 1648 or 1658 for the Ingelheim marriage?
The marriage that carried this moated Spessart castle to the Ingelheim family is dated 1648 in one tradition, a tidy match with the Peace of Westphalia, and 1658 in the Historisches Lexikon Bayerns. Ten years apart, both from credible sources. Where we need to mention it, we phrase the transfer without the year rather than lend a false precision to whichever number a reader might carry away.
Marienberg Fortress: founded in 1229 or around 1250?
Würzburg’s hilltop fortress is given a span of roughly 1250 to 1720 by the operator and 1229 to 1720 by the Historisches Lexikon Bayerns, which turns a neat “466 years, from 1253” into something less certain at the front end. We hedge the opening to “nearly five centuries” and “the mid-thirteenth century,” and we do not launder either endpoint into a hard figure.
The Wartburg ink stain: when did it vanish?
Luther is said to have thrown his inkwell at the devil in the Lutherstube, and for centuries visitors were shown the mark. The Wartburg-Stiftung itself published the monograph that calls the story a legend: Hilmar Schwarz, Die Legende vom Tintenfleck (1991). Physical ink appears only in the late seventeenth century, more than a hundred years after Luther’s death, and it survived because visitors scraped fragments off the wall as relics and the wall was re-inked for the next arrivals.
When the renewals stopped is where the record thins. Accounts based on the Wartburg literature put the stain’s disappearance at about the beginning of the twentieth century, and note a temporary renewal as a modern art action. We have seen more precise dates quoted for the last refreshing and for the earliest photograph showing a blank wall, and we could not confirm either against a source we could open, so we do not print them. Our pages say the wall has been blank since about 1900, which is what the record will carry.
Dover Castle: when was the Roman lighthouse built?
Three official records give three dates, and the most exact of them is more than half a century earlier than the operator’s. English Heritage, in its own exhibit on the pharos for Google Arts & Culture, says that “Around 2,000 years ago, in the early 2nd century AD, the Romans built a pharos, or lighthouse, here.” Later in the same exhibit it hedges: “The pharos was probably built on Castle Hill around this time …”, meaning the first half of the second century, when Dover became the base of the Roman fleet in the Channel. Historic England’s scheduling record for the castle, last amended in 1999, says “In around the 1st century AD a pair of lighthouses were constructed on the headlands flanking either side of the major Roman port of Dubris to help guide in cross-channel traffic.” Its separate Grade I listing of the pharos, first listed in 1974 and showing no amendment since, is the most exact of the three: “AD 46. Built under the Emperor Claudius.” It adds that the lighthouse guided the Roman fleet round to the port of Richborough, which neither of the others says.
So the most precise date sits in the least revised record. Our Dover Castle guide gives all three, each with its source, and says that no single date can honestly be stated. Its Quick Facts table gives the lighthouse no date at all.
Château de Caen: when was the Salle de l’Échiquier built?
Three credible accounts date the great hall at Caen, and they put it under two different kings. The castle’s own website, which is the Musée de Normandie’s, dates the hall to about 1098 and says it was probably built for receptions at the court of William Rufus, son and heir of William the Conqueror on the English throne. France’s national monuments record, the Mérimée database, says the stone keep and the hall were raised about 1120 by Henry I Beauclerc. And in the Bulletin Monumental in 1967, Francis Salet, reporting Michel de Boüard’s study of the building, wrote that there is every reason to think the hall was built in the first half of the twelfth century under Henry I, since the excavations had found no substructure of the Conqueror’s time there.
The split is one against two, and the one is the operator. We do not overrule the museum on its own building, and we do not follow it against the national record and the excavation evidence either. Our Château de Caen guide gives all three dates, each attributed, and calls the hall Romanesque, which is true on every reading. On the day it was published, its introduction called the hall twelfth-century and so quietly took Henry I’s side; that has since been corrected.
Measurements that describe different things
Château de Saumur: how high above the river does it stand?
Tourism material says the château stands “about 60 m above the Loire”; scholarship gives “about 40 m above the valley floor.” These are not two estimates of the same thing. Both are heights measured against a river or a valley, not an elevation above sea level, and a height quoted against a river is a different measurement from an altitude. No true metres-above-sea-level figure exists in any source we could find, so the guide describes the promontory in words and gives no number at all rather than dress a river-relative figure up as an altitude.
Wales: how many castles does the Welsh Government say there are?
Four different numbers, on three of its own websites, read on the same day. Cadw gives 427 on its castles page and over 600 on the page asking whether Wales is the castle capital of the world; on the same page as the 427 it says it looks after forty-four. wales.com says more than 400 at last count, and Visit Wales says over 641. Neither Cadw page carries a date. Its statutory schedule, which does have a rule, records 104 monuments whose site type is Castle. We print all of them with their sources rather than choosing one, because the spread is the finding: see the country with the most castles.
Is a tower house a castle?
Ireland says yes and Britain says beside. Ireland’s Sites and Monuments Record files Castle - tower house inside its castle group, 1,321 of them, and keeps a separate House - fortified house class as well, so the choice is considered rather than loose. Canmore, the FISH thesaurus and the Royal Commission on the Ancient and Historical Monuments of Wales all place TOWER HOUSE under FORTIFIED HOUSE, a term their vocabularies relate to castle and never place beneath it. Neither is wrong. The counts simply cannot be added, and a national total moves by hundreds depending on which filing you accept.
Vaux-le-Vicomte: how long is the garden?
Figures from roughly three to four kilometres circulate for the great perspective at Vaux-le-Vicomte. They measure the full estate and its approach. The designed perspective that André Le Nôtre actually composed for a single viewer, from the château to the Farnese Hercules, is about a kilometre and a half, with the Grand Canal running nearly a kilometre across it. We commit to the designed figures and are explicit that the larger numbers describe the whole domain, not the view.
Château de Blois: how many panels in the study?
Blois’s carved study, or studiolo, gets two numbers from the château itself, and they count different things. Its page on the royal apartments says that “The decor consists in 180 sculpted, painted and gilded oak panels that dissimulate four small cabinets, often with concealed doors, which contained small artifacts.” Its page on the visit, listing the château’s decorative motifs, counts “237 candelabras in the Studiolo.” Guidebooks have printed 237 as the number of panels since at least Hachette’s Guide du voyageur en France of 1895, which counted “237 panneaux d’ornementation différente,” and Michelin’s green guides were still printing it in the 2010s. Balzac, describing the room he calls both Catherine’s cabinet and her oratory, put its woodwork at “about a hundred and eighty oblong panels, one hundred of which still exist …” in Katharine Prescott Wormeley’s translation of Catherine de’ Medici.
None of these sources says how it counted, so they cannot tell us whether 180 and 237 describe the same panels. Until October 2026 our guide printed 237 as the panel count; it now describes the carved, painted and gilded oak panels and the four cupboards they hide without a number, as our valley-wide history essay already did.
Dover Castle: how tall is the Roman lighthouse?
English Heritage’s exhibit on the pharos says “The tower stands 15.8 metres high and is 12.2 metres wide at the base,” and “The Roman fabric survives to a height of 12.5 metres,” with the wall of a medieval belfry above it. Historic England’s scheduling record puts the same building at “approximately 19m in height” and goes further: “The architecture of the lighthouse suggests that it originally stood to a height of around 24m, but it has been extensively modified.”
Both standing figures are for the tower as it is today, and they are about three meters apart. Neither document says how, when or from what point its figure was measured, so the page cannot tell us whether the two describe different things. English Heritage mentions “a recent survey of the building itself” in the same exhibit, in a passage about the tower’s original profile, without saying whether its figures come from it. Our Dover Castle guide records both sets with their sources and does not average them.
Conwy’s town walls: 1.2 or 1.3 kilometers?
Two national bodies give the length of the town walls at Conwy, and their figures are a hundred meters apart. Cadw, which looks after the walls, gives 1.3 km twice: its scheduling record says they “measure approximately 1.3km in length and enclose an area approximately 22 acres,” and its page for the castle describes the town as “still protected by an unbroken 1,400-yard (1.3km) ring of town walls.” On Coflein, the Royal Commission’s record says: “The walls run for some 1.2km enclosing a roughly triangular 9.0ha area.”
Cadw’s own page for the walls uses a third form of words, saying they “run for three quarters of a mile, with 21 towers and three original gateways along their length.” Three quarters of a mile is about 1.2 km, the Royal Commission’s figure, while 1,400 yards is nearer four fifths of a mile. The two bodies agree on the area, since 22 acres and 9.0 hectares are close to the same thing, so the disagreement is over the line rather than the town it encloses. Neither record says where its measurement starts and stops, or whether the stretch the castle itself supplies is counted, though the scheduling record says the walls “incorporate Conwy Castle.” A surveyed length published with its end points would settle it. Our guide to Conwy Castle quotes the walls page’s three quarters of a mile and prints no decimal figure.
How many castles did the Welsh build?
Two Cadw documents answer, and their numbers do not match. Cadw’s page on Dolwyddelan says: “In fact, out of more than 470 fortresses across Wales, fewer than 40 were definitely built by native hands.” Its booklet on the castles of the Welsh lords and princes says: “Although relatively few (less than 30) of the surviving castles in Wales are formally recorded as being Welsh foundations, the original numbers must have been much higher.” Read quickly, they look like two attempts at one count. Read slowly, the two figures measure different things. The first counts fortresses that were definitely built by native hands, out of a total the page puts at more than 470. The second counts surviving castles formally recorded as Welsh foundations and says the original number must have been much higher; the booklet’s only total, on the same page, is a separate one for “the remains of up to 500 earthwork and stone castles …” Neither document lists the sites behind its figure, so neither can be checked against the other, and we print both with their wording rather than either as a bare count. For the totals themselves, see the Welsh Government’s four counts of all castles, above.
Attributions no one can pin down
The Très Riches Heures: who painted the Saumur page?
Château de Saumur appears in the September miniature of the Très Riches Heures du Duc de Berry, one of the most famous illuminated manuscripts in the world. The painter of that particular leaf is genuinely unsettled: the Limbourg brothers, Barthélemy d’Eyck working in the 1440s, and Jean Colombe who completed the manuscript in the 1480s have all been proposed. We attribute the manuscript at the level scholarship allows (made for Jean de Berry, begun by the Limbourgs 1412 to 1416, completed by Colombe 1485 to 1489, now Musée Condé, Chantilly, MS 65) and we name no painter for the September page itself.
Loches: who carved the effigy of Agnès Sorel?
At Loches, the recumbent effigy of Agnès Sorel is dated to the fifteenth century by the French national heritage inventory, which names no sculptor. A tradition dating from an 1885 article credits the sculptor Jacques Morel, other scholars have rejected him, and the attribution remains debated. We describe it as “a fifteenth-century recumbent effigy, its sculptor still debated,” and name no candidate as fact.
Chambord: who designed it?
The Domaine national de Chambord says two things on the same page, and both are true. It reads the double staircase, the centre-plan keep and the terrace drainage as evidence of Leonardo da Vinci’s hand behind François I’s project. It also states plainly that the architect’s identity “has remained an enigma”, and that no plan and no text contemporary with the building gives a name. It even supplies the reason for the silence: the site’s records were dispersed or destroyed at the end of the eighteenth century.
Names circulate anyway, Domenico da Cortona and Pierre Nepveu among them, and none of them is established. Our guide and our film report the attribution and the absence side by side, and assert no architect. A five-hundred-year-old building that half of Europe has photographed has no documented author, and that absence is a finding rather than a gap waiting for a confident answer.
Les Fâcheux: who wrote the music?
Molière’s Les Fâcheux, the first comédie-ballet, was premiered at Vaux-le-Vicomte during Fouquet’s fête on 17 August 1661. Who composed it is not agreed. The Vaux-le-Vicomte estate says the play was “created in collaboration with choreographer Beauchamp and composer Lully”. The Philharmonie de Paris, writing about the genre, says Pierre Beauchamp composed the bulk of the score and that Lully contributed a single piece, the courante for Lysandre.
Neither is a careless source, and the disagreement is about emphasis as much as fact: Lully’s name carries the work in catalogue shorthand, while the detailed account of the score credits Beauchamp. We went back and forth on this one and landed on naming both men and their roles without adjudicating the score, because the hub block is 200 words and this needs more room than that. If you can settle it from a seventeenth-century source, we would like to hear from you.
Dolforwyn Castle: who took it in 1277?
Llywelyn ap Gruffudd built Dolforwyn in 1273, on Cadw’s account, and lost it four years later. Who took it depends on which record you open. Cadw’s booklet on the castles of the Welsh lords and princes says “Llywelyn had evidently completed the castle’s defences before it was besieged and captured by Edward I in 1277.” Cadw’s own page for the castle names the Marcher lord Roger Mortimer alone. The chronicle of the princes, Brut y Tywysogion, in the Rolls Series translation of 1860, names two besiegers and a cause: “the earl of Lincoln and Roger Mortimer besieged the castle of Dolvorwyn, and at the end of a fortnight they obtained it, through want of water.”
Two of the three accounts are Cadw’s, and they do not agree with each other. The booklet’s Edward I may be shorthand for the king’s campaign rather than a statement about who stood outside the walls, but the booklet does not say so, and choosing that reading for it would be a guess. Our article on the castles of the Welsh princes prints all three, each attributed, and lets the disagreement stand.
Pembroke Castle: which Montgomery began it?
Cadw and the castle’s operator both put the start of Pembroke in or around 1093, and they part company over which Montgomery began it. Cadw’s scheduling record for the castle says the site was “chosen early in the first Norman incursions into south-west Wales, when the castle was founded by Roger of Montgomery in 1093, the castle was never to fall to the Welsh.” His son is the builder on the castle’s own history page: “Pembroke Castle has a long and fascinating history, for it was around 1093 that Arnulf de Montgomery built the small inner bailey standing at the end of the promontory.” Gerald of Wales names Arnulf too, in Richard Colt Hoare’s translation of the Itinerary of Archbishop Baldwin through Wales: “Arnulph de Montgomery, in the reign of king Henry I., erected here a slender fortress with stakes and turf, which, on returning to England, he consigned to the care of Giraldus de Windesor, his constable and lieutenant-general, a worthy and discreet man.”
Gerald’s reign does not sit well with 1093, and Hoare’s own note says as much: it has the fortress strong enough to have “resisted the hostile attack of Cadwgan ap Bleddyn in 1092, and of several lords of North Wales, in 1094.” The first of those dates falls a year before the 1093 that Cadw and the operator give, so Gerald and his translator count as witnesses to a name rather than to a date. The same note calls Arnulf the “younger son of Roger de Montgomery.” Brut y Tywysogion, the chronicle of the princes, names the son as well: in the Rolls Series translation, its entry for 1100 calls Ernulf, the brother of Robert, earl of Shrewsbury, “the person who had obtained Dyved for his share, by ballot, and who magnificently built the castle of Pembroke.” Three accounts name the son as the builder, and the scheduling record names the father as the founder. A reader may suspect that both are true at once, the father as lord and the son as builder, but none of the records we read says so. Our guide to the castles of Wales says Pembroke “was founded by the Normans in 1093” and names no founder.
Castell Dinas Brân: who built the stone castle?
Three documents from two Welsh public bodies describe the stone castle on the hill above Llangollen, and each names its builder differently. Cadw’s scheduling record names a family: “A medieval stone-built castle was later constructed during the 1260s by the lords of Powys Fadog, in the northwest corner of the hillfort.” Cadw’s booklet on the castles of the Welsh lords and princes names a man: “An iconic ruin on a towering site in the Dee valley overlooking Llangollen, built by Madog ap Llywelyn of Powys Fadog probably during the 1260s.” On Coflein, the Royal Commission’s record names another: “It was probably Gruffudd ap Madog, son of the founder of nearby Valle Crucis Abbey, who rebuilt the castle in stone c.1270, but much like its predecessors, this third and last castle did not last long.”
Dates pull against the names as well. Brut y Tywysogion, in the Rolls Series translation, records under 1269 that on “the seventh day of the month of December, Gruffudd, son of Madog, lord of Maelor, and Madog the Little, his brother died, and were buried at Llanegwestl.” Valle Crucis was also known in the Middle Ages as the abbey of Llanegwestl, and the same chronicle has Madog, son of Gruffudd Maelor, founding the monastery there, so its Gruffudd, son of Madog, is the Royal Commission’s man, dead before the “c.1270” the record gives for his rebuilding. The Royal Commission’s record also carries an older entry, from 2007, which says the castle “is thought to have been built in about 1270, if not rather earlier,” and so leaves room for the scheduling record’s 1260s. The scheduling record’s family does not rule out either man, but the booklet’s Madog appears in neither of the others, and none of the three documents sets out the evidence behind its name. A document naming the man who ordered the work would settle it; none of the three cites one. Our article on the castles of the Welsh princes prints the scheduling record’s “lords of Powys Fadog” and the Royal Commission’s Gruffudd side by side, with the chronicle’s date of his death beside them.
Hohenschwangau: who built the castle of 1538, and how is his name spelled?
Two Bavarian state records describe the same campaign of 1538 to 1547 and name its builder differently. Haus der Bayerischen Geschichte writes Lucio de Spari and calls him Neapolitan, offering the Castel Nuovo as a probable model for the box-like residential block he raised. The Bavarian heritage office, in its statutory listing, writes Lucio di Spazzi and gives no birthplace at all. Circulating accounts add a third version, making him a Lombard from Lanzo d’Intelvi with a spell at the Innsbruck Hofburg, and that one has no state carrier behind it.
So our guide names the man, attributes the Neapolitan origin to the institution that asserts it rather than to itself, prints both spellings beside the record that uses each, and states no birthplace of its own. The Innsbruck claim we leave out entirely: an unsourced detail repeated often enough still has nothing under it.
When a source contradicts itself
Burg Eltz: eighty rooms or more than a hundred?
The same operator contradicts itself across languages. The German pages of burg-eltz.de give the castle up to eighty rooms; the English pages of the same site say more than a hundred. That gap is the likely origin of the figures that circulate elsewhere. Twelve rooms are shown on the guided tour, which is the number a visitor can actually check, so that is the one our guide leads with, and we report the discrepancy rather than choosing a side.
Burg Eltz: is the Banner Hall of 1470 or about 1480?
A third self-contradiction on the same site. The building chronology on burg-eltz.de dates the Fahnensaal, the Banner Hall with its late-Gothic net vault, to 1470; the guided-tour description on the same site calls it the Banner Hall of around 1480. Nothing we could find settles which, and the difference is not one a visitor can see, so our guide gives the range rather than picking the page we happened to read first.
Burg Eltz: 1835 or 1838 for the Diezler view?
The most interesting conflicts are the ones a source has with itself. On the official Burg Eltz website, the page describing Anton Diezler’s painting of the castle gives two different years: the image filename says 1835, while the caption text beside it says 1838. We ship 1838, the operator’s considered caption value, and we do not “correct” it to 1835, because chasing the filename would only be chasing a different guess. It stays logged as an open question that a third source would settle.
Ewloe Castle: which Llywelyn built it?
Cadw looks after Ewloe, and three of its own documents do not agree on who built it. Its page for the castle says that, for lack of records from the period, “the castle’s history is a little murky,” and then reads it in two stages: “The characteristically Welsh D-shaped stone tower was probably built by Llywelyn ap Iorwerth (Llywelyn the Great) some time after 1210, with the curtain walls and circular western tower being added by Llywelyn ap Gruffydd (Llywelyn the Last) nearly 60 years later.” Its booklet on the castles of the Welsh lords and princes reports “much debate about whether Ewloe was begun by Llywelyn ap Iorwerth in the 1220-30s or Llywelyn ap Gruffudd after 1257,” and leans the same way: “the evidence now seems to favour Llywelyn ap Iorwerth as the first builder on the site.” Its scheduling record for the monument takes the other side, describing “a medieval castle built around 1257 by Llywelyn ap Gruffudd.”
The booklet also names the one document that refers to the castle explicitly: a report in the Chester Plea Rolls of 1311, which suggests that Llywelyn ap Gruffudd restored the castle after recovering the manor of Ewloe from the English in late 1256. Our article on the castles of the Welsh princes quotes all three Cadw readings, each beside the document that makes it, and names no builder in its own voice. Where the operator has not settled its own account, we are not going to settle it on Cadw’s behalf.
Did the Welsh princes build in stone before the thirteenth century?
Cadw answers no on one page and yes in one of its booklets. Its page on the castles of Wales says the princes’ earliest castles “were like those of the Norman invaders in design, and were also of earth and timber,” and goes on: “But in the thirteenth century the Welsh princes began to build their castles in stone and the design was very different.” Its booklet on the castles of the Welsh lords and princes says of Cardigan: “Recorded as having been built in stone by the Lord Rhys in 1171 and the site of the first Eisteddfod.” It adds that most of the masonry visible there now is likely to be the work of William Marshal or the English Crown in the mid to late thirteenth century, “but excavations have revealed the remains of possible earlier walls.” The chronicle of the princes, Brut y Tywysogion, agrees with the booklet: in the Rolls Series translation, the lord Rhys built the castle of Aberteivi, Cardigan, “with stone and mortar …” Gerald of Wales, crossing the estuaries of north-west Wales in 1188, passed the place “where two stone castles have newly been erected …” One of them, he writes, was “Carn Madryn, the property of the sons of Owen …” and the booklet places that castle on the summit of Carn Fadryn and gives it to the sons of Owain Gwynedd.
A second Cadw page draws the line where the evidence allows it: “It wasn’t until the 13th century that Welsh princes began to construct stone castles as mighty as those raised by the Marcher lords on the border.” The great stone castles of the princes do belong to the thirteenth century. Stone itself came earlier, and Cadw’s own booklet records it.
Dinefwr Castle: when was the north-west tower built?
Cadw looks after Dinefwr, and two of its own documents disagree about when the castle’s smaller round tower was built. Its scheduling record for the monument describes, besides “the great round keep on the east,” another round tower, and dates it in a single sentence: “The circular three-storey tower on the north was probably constructed after 1277.” Its ground plan of the castle labels that tower “North-West Tower” and shades most of its walls in the color the key gives to “Early/Mid-Thirteenth Century: Phase 2,” a phase set apart from the band the same key calls “Late Thirteenth/Early Fourteenth Century.” On the plan, the later color is confined to the tower’s inner side, where it joins the hall.
Both documents are describing one building. The record goes on to place its tower beside “the adjacent rectangular hall,” and on the plan the north-west tower is the round tower joined to the hall. Coflein, the Royal Commission’s online record, which names the Cadw guidebook by Rees and Caple as its source, mentions “a smaller round tower at the northern angle” and gives it no date. Neither Cadw document gives its reasons, and no source we could reach dates the tower on evidence of its own, so we report both dates and settle neither. Our article on the castles of the Welsh princes draws only Dinefwr’s round keep and does not date this tower.
When the specialists overrule the operator or a reference book
Burg Eltz: was the first cannon north of the Alps fired here?
Burg Eltz states in its own history that the first secured cannon attack north of the Alps took place at the castle in 1331, during the siege by Archbishop Balduin of Trier. The claim rests on a group of gun-arrows, partly of wood, long kept at the castle. When Olaf Wagener and Achim Schmidt published their building-archaeological study of the siege castle in the Trierer Zeitschrift in 2010 and 2011, they examined that evidence and concluded it cannot carry the claim: none of the arrows was recovered during the clearance of the west zwinger where they were said to have been found, they were already at the castle by the early twentieth century with their origin unrecorded, one is a modern replica made by a former castellan, and a nineteenth-century purchase cannot be ruled out. Their verdict is that the finds should not be cited as evidence at all.
This is the hardest kind of conflict for us, because the operator is normally our best source for its own castle, and here the peer-reviewed specialists contradict it. We follow the specialists: our Eltz guide reports the castle’s claim, reports the study, and declines the superlative. We would rather lose a striking fact than repeat one the evidence will not support.
Burg Eltz: has one family really held it since 1157?
Burg Eltz is famous for an unbroken line of family ownership, and three institutions state it plainly. The castle’s own history says the building is today in the hands of the 34th generation of the lords and counts of and at Eltz, without interruption since the middle of the twelfth century. Rhineland-Palatinate’s Generaldirektion Kulturelles Erbe describes furnishings in which more than eight centuries of continuity of the counts at Eltz as owners are reflected. The Deutsche Burgenvereinigung, awarding the family its Grosser Denkmalpreis in 2024, says the castle is already in the 34th generation of family ownership.
Alexander Thon, writing on the Rhineland regional-history portal of the Landschaftsverband Rheinland, qualifies what that continuity can be shown to mean at its earliest end. On his account Eltz was an imperial castle for close to two centuries after its first appearance in the records, garrisoned by unfree imperial ministerials who bore the name Eltz from the place of their service rather than as a family name in the modern sense, and a non-noble Eltz ministerial line is demonstrable only from 1268, at which point it appears to have received Eltz from the Roman-German kings as a hereditary fief.
The two accounts are not in fact describing the same thing. Neither has the rock changing hands; the disagreement is about what the surviving charters can be made to prove about the men who held it before 1268, and about whether a name taken from an office is the same as a family name. We give both. Our Eltz guide opens on the continuity as its thesis, attributes the dated version to the bodies that make it, and sets Thon’s qualification beside it in the history section. The opening sentence is deliberately the one place the claim still stands in our own voice.
Burg Eltz: how far away was the siege castle?
Trutzeltz, the siege castle Archbishop Balduin of Trier raised above Burg Eltz during the Eltz Feud, sits on the ridge to the north, and the distance between the two is given differently by our two best sources. The KuLaDig regional heritage register places it about 170 meters north of the approach to Burg Eltz, a figure it takes from a 2006 study by Eulenstein. Olaf Wagener and Achim Schmidt, in the building-archaeological survey of the siege works published in the Trierer Zeitschrift, put the distance of the two fortifications from the besieged castle at between 200 and 240 meters.
The gap is large enough to matter for anyone trying to picture the siege, and neither figure can be dismissed. It may not be a disagreement at all: KuLaDig measures 170 meters north of the castle’s approach, while Wagener and Schmidt measure from the besieged castle itself, so the two are counting from different points. We print no horizontal distance in the Eltz guide’s own prose, carry both figures with their datums in the source note, and record the divergence here. The height does not settle it either. Wagener and Schmidt measure just under forty meters from platform to platform, while KuLaDig puts Burg Eltz at 129 and Trutzeltz at 182 meters above sea level and calls the difference around fifty. Our guide prints the scholarly platform figure, because a height derived by subtracting two sea-level readings is the wrong datum for how far one castle rises above another.
Vaux-le-Vicomte: was Le Nôtre “Premier Jardinier du Roi”?
The Vaux-le-Vicomte estate’s account of André Le Nôtre says he was “later named Premier Jardinier du Roi”. At Versailles, which holds the deeper record of his career, he is recorded in royal service from 1635 and head gardener of the Tuileries from 1637, and gives 1657 for the controller-generalship of the king’s buildings. The specific title the Vaux page uses is one the scholarship does not record him holding at all.
The difference matters because it changes the story. If Le Nôtre was unknown before Vaux, Louis XIV discovered him there; if he had held royal appointments for two decades, Fouquet hired established men and the king took the team. We follow the specialists and avoid the title: our pages say Vaux was the first project on which Le Vau, Le Nôtre and Le Brun worked together, which is what the Vaux estate itself says and what the record supports.
Dinefwr Castle: who surrendered it in 1277?
The Dictionary of Welsh Biography gives the castle to the wrong Rhys. Its entry on Rhys ap Maredudd says: “In 1277 he had submitted to Edward, surrendering the castle of Dinefwr, but being allowed to retain Dryslwyn.” It follows J. E. Morris, whose Welsh Wars of Edward I (1901) has Edward’s captain in West Wales bring Rhys ap Maredudd “to submit and to surrender under terms his castle of Dynevor on that river.” Morris rests on the agreement of 11 April 1277, printed in Rymer’s Foedera, and the Latin says something else: the king, once he holds Dinefwr (“cum in praedictis castro & terris habuerit potestatem …”), will do Rhys justice and restore it to him according to his right. On that day the castle was not the king’s to receive, and Rhys ap Maredudd was claiming it, not giving it up.
The man who held it was his kinsman Rhys Wyndod. In Huw Pryce’s edition of the acts of the Welsh rulers, the summary of Rhys ap Maredudd’s own petition says Payn de Chaworth and his companions promised him the castle on the king’s behalf “if they could conquer it from Rhys Fychan [i.e. Rhys Wyndod] and his men …” Pryce also identifies Wyndod as the “Reso Parvo de Estrad Tywy” of the grievances sent to Archbishop Pecham in 1282, and the first of them, printed in the archbishop’s register, opens with his having given the king “castrum suum apud Dynewr …,” his own castle at Dinefwr. Ralph Griffiths dates Wyndod’s submission to 24 April 1277 and writes that “on 5 June, his ancestral fortress of Dinefwr was seized in the king’s name.” Rhys ap Maredudd did give up Dinefwr to Edward in the end, but six years later and only as a quitclaim of his right, at Acton Burnell on 16 October 1283. Our guide to the castles of Wales and our article on the castles of the Welsh princes now name Rhys Wyndod.
Questions the evidence cannot settle
Some conflicts are not between two sources but inside the evidence itself. These are the cases where we decline to give an answer, and say so on the page rather than pick the tidier version.
Ludwig II of Bavaria: how did he die?
On 13 June 1886, the day after he was taken into custody at Neuschwanstein and interned at Berg, Ludwig II was found dead in Lake Starnberg together with the psychiatrist von Gudden. The Bavarian Palace Administration, the operator of his palaces, does not offer a cause: its own account says he died in mysterious circumstances.
The medical literature is still arguing. Steinberg and Falkai reopened the case in the European Archives of Psychiatry and Clinical Neuroscience in 2020, and their reading is contested by other specialists. Drowning, a struggle, a heart attack and a shooting have all been argued from the same thin evidence, and no autopsy settles it.
So our guide and our film state what is documented and stop. That is less satisfying than a conclusion, and it is the honest position: where the specialists have not settled a question, we are not going to settle it for them.
Hohenschwangau: what was the round tower of 1363?
A round tower recorded in 1363 as the Synweln Turm stood on this site before the castle the heritage register first mentions in 1397, and two Bavarian state institutions read the same documents and arrive at different buildings. The castle database of Haus der Bayerischen Geschichte, written by the archaeologist Joachim Zeune, makes that tower the direct predecessor of today’s palace, appearing from 1476 as Burg Schwanstain, and describes Hohenschwangau as built towards the middle of the fourteenth century as a round residential tower. The Historisches Lexikon Bayerns instead lists Schwanstein and the round Scheiblingsturm as two separate castles among the family’s six.
A state institution’s castle specialist against the state library’s encyclopedia is not a case of one good source and one bad one, and neither publishes the reasoning that would let a reader choose. Our guide reports both readings with both institutions named. Worth noting in passing that sinwel is Middle High German for round, so the tower’s two recorded names both mean round tower and neither of them says what it was for.
The Schwangau swan: black or silver?
The arms that gave both castles in this valley their names disagree with themselves across six centuries. In the Codex Manesse, the Grosse Heidelberger Liederhandschrift, the minnesinger Hiltbold von Schwangau is shown with a black swan on a red field. The official blazon, adopted as the municipality’s arms in 1953, gives a silver swan armed black with wings spread, on red.
Two credible records, two tinctures, and the older witness is the manuscript rather than the modern blazon. This is not a case where one supersedes the other: a medieval song manuscript and a twentieth-century municipal grant are different kinds of document doing different jobs. Our guide gives both and says which is older, and its captions describe the swan without committing to a colour the sources divide on.
Dolbadarn Castle: which came first, the walls or the tower?
Dolbadarn is two kinds of building on one rock: a round tower of mortared masonry, and an enclosure whose curtain wall Cadw’s scheduling record describes as unmortared. C. A. Ralegh Radford’s Ministry of Works guide notes that the castle “is mentioned in no document earlier than the English conquest,” so the order has to be read from the stones, and four official accounts read it three ways.
Cadw’s scheduling record calls the unmortared curtain “the earliest work” and the round tower “a later addition dating from the 1230s.” Radford agreed on the order: “The masonry is of two dates, both older than 1284,” and the curtain’s plan and character “suggest a date in the second half of the twelfth or the early years of the thirteenth century.” The Royal Commission’s 1960 inventory of Caernarvonshire allowed that the courtyard “could conceivably be earlier, possibly of the late 12th century, but it seems too slight a work to have been designed without the round tower and is more likely to be contemporary with it or nearly so, their dissimilarity perhaps being due to division of the work between local men and a master mason from another region.” Cadw’s listing record reverses the order altogether: “The curtain wall and outbuildings, of which only fragmentary ruins survive, post-date the keep and may have been added by the English.”
Two of the four are Cadw’s own records, and they disagree. Coflein, the Royal Commission’s online record, repeats the scheduling record’s account almost word for word and names its source, Richard Avent’s 2004 Cadw guidebook, so we count the two as one reading, not two. Our Dolbadarn guide sets out all four readings with their sources and says they cannot all be right. It does not pick one, because nothing we could reach decides between them.
Did the Welsh princes take readily to the motte?
Cadw and the castle historian D. J. Cathcart King read the same early period in opposite ways. Cadw’s page on Dolwyddelan says: “Early Welsh princes didn’t readily adopt the Norman practice of building castles.” It gives a reason in the next sentence: “They preferred a more nomadic lifestyle moving between undefended ‘llysoedd’, or courts, spread across their lands.” King, in The Castle in England and Wales (1988), starts from the princes’ enemies: “Welsh princes, however, had many less terrible potential enemies, English and perhaps more particularly Welsh, who could be held off by well-sited strongholds.” His conclusion runs the other way from Cadw’s: “Thus we find the motte-castle, alien institution though it certainly was, readily accepted by the native rulers; as early as 1111 Cadwgan ap Bleddyn was reconnoitring a site for one near Welshpool when he was assassinated.”
The two are not quite answering the same question. Cadw is speaking of castle-building in general and of how the princes lived; King is speaking of one form of castle and of the rulers who chose it. Neither passage counts the princes’ mottes. Our article on the castles of the Welsh princes prints both.
Found something we got wrong?
If you can settle one of these with a primary source, or you have spotted an error anywhere else on the site, we genuinely want to hear it. Write to info@stonekeepatlas.com or use the contact page, and point us at the source if you have one. Corrections are made to the article itself, and the more interesting conflicts end up here.