The Paper Siege: sequential parcel capture and permission-framing
The leading family working hypothesis is that separate land transactions, public-body takings, possession proceedings, licences, registrations and title consolidation progressively converted the Williams family’s practical occupation and asserted rights into a subordinate or permissive status on paper, while corporate and public bodies consolidated control over different parts of the estate. The model tests sequential parcel capture or title-scope expansion. It does not presently establish a single coordinated plan, fraud, forgery, unlawful appropriation or violence.
Status: Leading working hypothesis — unproven; not a finding of fact or law
How to read this: the model starts with a leading hypothesis so it can be tested. It does not treat a hypothesis as a fact. The court outcome, source gaps, alternative explanations and disconfirming tests remain visible alongside it.
Scope and essential corrections
The c.1841 tithe apportionment records John Williams as occupier of 36 numbered plots. Occupier is not owner. The current catalogue describes 33 entries with Bute as landowner and three with Revd James Evans / glebe; this is a strong footprint of occupation, not proof that Williams owned all 36 or that all 36 were later Parcel A or B.
Separate sub-hypothesis: the Williams family may have held a proprietary, beneficial, customary, leasehold or other inheritable interest across some or all of the 36 plots occupied by John Williams. The c.1841 tithe is not evidence that he owned them outright: its owner column reportedly attributes 33 plots to Bute and three to Revd James Evans / glebe. The hypothesis therefore concerns whether the family held an interest not captured by the freehold owner column, or whether later records obscured a prior interest. It is directly weakened if the original schedule and source deeds show ordinary occupation under Bute / Evans without any separate proprietary interest.
The c.1841 tithe parcels, the provisional later Parcels A, B and C, and the A1/A2/A3 sub-parcels are different mapping systems. They must not be treated as identical until each historic plot is reconciled against original maps, deeds and plans.
1939 agreement
The 1939 WGR–John Williams agreement is presently associated with the fields / provisional Parcel B. Its plan does not show the farmhouse parcel. Its complete wording, parcel schedule, original signature and provenance remain material questions. The family’s explanation that this is why Mary did not contest the taking of Parcel B is recorded as a hypothesis, not as an established statement by Mary.
Parcel A
Parcel A is the separate house / farmhouse / garden / yard / woodland claim as currently modelled. The exact boundaries and the relationship between the 1938 title, 1939 letting and later first registrations remain unresolved.
Parcel C
The southern green / VG41 area is kept separate as provisional Parcel C; it must not be silently folded into Parcel A.
If the leading hypothesis were true, it would explain…
If the sequential-capture theory were true, the 1841 tithe would show that the Williams family’s occupation was broad and longstanding, while later estate and company records progressively narrowed the land treated as theirs. It would not, by itself, prove freehold ownership.
If the 1939 agreement concerned only Parcel B, it would explain how one distinct field tenancy could later be used in argument about the family’s general status without proving that the farmhouse, garden or woodland were included in that letting.
If John Williams signed the 1939 WGR agreement and it genuinely covered only the fields now modelled as Parcel B, it could explain why Mary regarded that taking as a separate matter and concentrated her own ownership claim on the farmhouse, garden and woodland (provisional Parcel A). But the 1987 Court of Appeal records a separate procedural explanation: the appeal was limited to the farmhouse and garden because the possessory-title case for the rest of the farm had failed on findings that the alleged acts of possession were too tenuous and equivocal. The 1939 agreement must therefore be tested as one possible explanation, not assumed to be the reason Parcel B was left uncontested.
If the signature on the 1939 agreement were not John Williams’s, or if the document had been altered or misattributed, it could explain why a disputed paper instrument became a hinge in the later account of permissive occupation. That possibility requires the original document, chain of custody and proper handwriting examination; a visual mismatch alone does not prove forgery.
If corporate or public bodies occupied, used or asserted rights over land outside the parcels conveyed or taken to them, it would explain the family’s 1958 complaint that land had been wrongfully appropriated and the later appearance of a split between practical occupation and official title. Each site and taking would still need its own plan and instrument.
If successive proceedings and offers repeatedly recast Mary’s occupation as tenancy, licence or occupation subject to a possession order, they could have prevented her continued presence from maturing into a limitation defence. The recurrence would explain the legal mechanism; it would not alone prove that the actors coordinated it or acted dishonestly.
If the 31 October 1974 letters were deployed after Mary pleaded limitation as a way to change her legal status without a negotiated tenancy or signed acceptance, it would explain the pivotal shift in BP’s case. The Court of Appeal expressly said it would be artificial to say Mary had accepted the letters; it nevertheless held that they unilaterally granted a lifetime licence and that, in the absence of repudiation, BP had to treat her as a licensee. A deliberate strategy is not established, but the legal effect and the timing are documented.
If the 1975 conveyance transferred only a defined farmhouse plot and other specified land, while retaining other land, it would explain the importance of comparing its actual plan and parcels with the 1969 root transaction and later registrations rather than assuming that every part of the estate moved together.
If the 1982–83 first-registration applications described BP as in possession or failed to identify Mary’s occupation or competing claim accurately, and if effective notice was not served, that could explain how a registered title was created or completed without a meaningful opportunity for her to object. The actual application forms, requisitions, notices, service records and correspondence must establish whether this happened.
If the later amalgamation and possession proceedings relied on an over-broad parcel description, they could have entrenched a title claim that originated in an earlier boundary or conveyancing error. The 1987 court decision remains legally significant for the matters it decided; it is not a substitute for independently testing every underlying instrument and parcel.
If the development sequence was connected to the acquisition of control, the 1989 outline application (outline permission 13 March 1990), William Buckler’s death-date discrepancy (family chronology: 29 May 1990; probable probate-calendar match for William Beverley Buckler: 29 April 1990, EV-0274) and the 3 September 1992 reserved-matters approval would be dates to reconcile. Both candidate dates fall between the planning milestones, but neither establishes cause, motive, knowledge or homicide.
If records were deliberately suppressed or destroyed, the reported loss of papers from the farmhouse and the reported Neath fire might form part of a concealment mechanism. At present these are separate leads: no established link proves a common actor, purpose, arson or deliberate destruction.
If the archaeology were being used to conceal an earlier secret, the excavation, treatment of human remains and documentary handling would need to show a deliberate connection to the title dispute. The number of burials, damaged skulls, a cellar or a possible former court does not establish prisoners, executions, a burial secret or a state operation.
If BP’s spokesman in 1988 described Mary’s death as the point when “it became clear her son was not going to leave” [EV-0088], it would explain how BP publicly framed the son’s occupation as a new phase after the personal licence offered to Mary ended. That statement is relevant to the chronology and corporate narrative; it does not show that BP caused her death, knew of her illness, or arranged anything.
Mechanism, stage by stage
Each stage states the proposition, what it would explain if true, the alternative and the artifacts to test. Dates do not establish motive.
M01 · c.1841–1928
Occupation footprint and possible change in tenure
Map John Williams’s 36 tithe plots, the 1870s rate-book entries, the 1916 yearly tenancy and the reported end of house-parcel rent in 1928. Test whether the family’s practical occupation or a separate interest in the house parcel was wider than the later corporate title description. Separately test the family’s broader 36-plot ownership theory: identify the nature of any claimed interest (freehold, leasehold, customary, equitable or beneficial) and reconcile it with the tithe owner column, which presently names Bute / Revd James Evans rather than John Williams.
If true: A broad occupation history would precede the later parcel-specific legal mechanisms. If the family held an inheritable or beneficial interest not recorded as freehold ownership in the tithe schedule, it could explain why later records show occupation but fail to carry that interest forward; the type and extent of the interest would need to be identified.
Alternative / evidence against: The tithe records occupation and named landowners; the 1916 tenancy is recited in the 1987 judgment and the original is not held; the 1928 rent account is family testimony. A tenancy can explain occupation without ownership. The tithe’s owner column is a direct counter-indication to outright Williams freehold ownership of all 36 plots; occupation alone can arise from tenancy or other permission.
Compare the 25 December 1938 conveyance and plan with the 18 May 1939 agreement and plan. Establish exactly which parcels WGR acquired, which it let to John Williams, and whether the farmhouse parcel was excluded from the 1939 letting.
If true: A narrow Parcel B letting could have been used later as a paper explanation for occupation without governing Parcel A.
Alternative / evidence against: The 1938 deed and plan are not yet independently inspected in full; a 1939 plan defining the letting does not, by itself, prove what the 1938 conveyance included. The Court of Appeal’s 1987 judgment says the possession order for the rest of the farm was not challenged on appeal because the pleaded possessory-title case failed on findings of fact that the acts relied upon were too tenuous and equivocal. That is a distinct judicial explanation for the limited appeal; it does not establish Mary’s personal reason for not contesting each earlier taking.
Reconstruct the 1949 surrender and re-grant, 1952 decision not to proceed with a letting, 1955 possession action, 1958 Crichel Down complaint, 1959 and 1965 tenancy offers, 1961 woodland rights and 1962 farmhouse possession order. Assign every event to a mapped parcel.
If true: A repeated sequence of taking, offer, refusal and possession proceedings could explain how separate land interests became layered and how Mary’s claim was narrowed or procedurally disadvantaged.
Alternative / evidence against: The 1987 judgment is the source for several earlier events; repeated recitals are not independent corroboration. Some actions may concern Parcel B, common land or woodland, not the farmhouse. The 1987 judgment records that possession of the farm’s other parts had been taken in July 1955, while the farmhouse and garden remained occupied; it says the appeal later concerned only the farmhouse and garden because the other possessory acts were too tenuous and equivocal. The parcel mapping must reconcile that judicial account with the family’s 1939-agreement explanation.
BP acquisition, Mary’s plea and the 1974 licence letters
Compare the 1969 WGR–BP Pension Trustconveyance, the 1971 abstract, the 1974 pleadings, the two 31 October 1974 letters and the 23 May 1975 BP Pension Trust–BP Properties conveyance. Distinguish claimant, legal owner, prospective purchaser, licensor and the land each instrument identifies.
If true: The letters could have been a decisive paper mechanism for changing the legal character of Mary’s occupation immediately after she pleaded limitation. The Court of Appeal did not find express acceptance; it held that the letters unilaterally granted a licence and, because Mary did not repudiate it, BP had to treat her as a licensee. If the letters were deliberately selected for that legal effect, they would explain the shift from a contested ownership/limitation plea to a licence-based account of occupation.
Alternative / evidence against: The Court of Appeal rejected the appeal and its reasoning is the operative legal outcome. Dillon LJ said it would be artificial to say Mary accepted the terms, but held that absent repudiation BP had to treat her as a licensee. The judgment does not establish that the letters were fabricated, that the strategy was pre-planned, or that the 1975 conveyance itself evaded limitation.
First registration, notice and title consolidation
Read the complete WA231076 and WA240304 files, including applications, statements of possession, lodged instruments, requisitions, replies, notices, service evidence and register entries. Then compare the 1987 amalgamation plan with the separate parcel histories. Include the 8 April 1983 D23 letter and April–May 1983 Registry correspondence to establish what the Registry was told after Mary’s death.
If true: A materially incomplete or inaccurate description of occupation, title scope or notice could explain why Mary’s competing position was not resolved before registration and why the later title became harder to challenge.
Alternative / evidence against: The fact that the applications were made while Mary was alive and that later correspondence followed her death does not establish what notice was served, whether it was required, whether it was effective or whether any appeal period was missed.
Put the 1988 eviction and demolition, the 1989 outline application, 13 March 1990 outline permission, 1992 reserved matters, archaeological correspondence and 1994 excavation into one dated chain, while keeping title, planning and archaeology as separate legal systems. The current research correction treats the 1993–94 excavation as recording an early medieval cemetery; the reported burial count and any Roman-villa connection remain unconfirmed in the catalogue and must not be treated as established.
If true: A completed title claim could have enabled demolition and redevelopment; unresolved archaeological and planning records could show how the physical site was transformed and what was documented or omitted.
Alternative / evidence against: Planning permission and archaeological work do not prove a title conspiracy. Burial counts and trauma patterns require archaeological context and cannot identify perpetrators or motives by themselves. A current fact-check corrects earlier claims of 1,026 Early Christian burials and a Roman villa complex: the cemetery is recorded, but the burial count and villa link are not yet confirmed. A cemetery is not evidence of prisoners, executions or concealment.
Build a person-by-person chronology for John Williams, Frederick Buckler, Mary Williams and William ‘Billy’ Buckler. Use original civil registration, probate, inquest, medical, police and contemporaneous correspondence where lawfully available.
If true: A proven pattern of threats, suspicious interventions or deliberate document removal could support a separate suppression hypothesis.
Alternative / evidence against: Deaths near legal or planning events do not establish homicide. John Williams’s death date remains unestablished (the family research records him alive and very ill in the early 1950s); Frederick Buckler’s date also remains unverified. For William Buckler / William Beverley Buckler, the family chronology says 29 May 1990, while a 1991 probate-calendar entry is a probable match dated 29 April 1990 (EV-0274); identity and date are not yet reconciled against civil and probate records, so no timeline is drawn from it inference. Mary’s death index and the 1988 BP spokesman’s reported wording are relevant context, not evidence of a caused death.
Unresolved; probable name match, not yet authenticated as the same person. Both dates fall after outline permission on 13 March 1990 and before reserved-matters approval on 3 September 1992. Do not infer causation from this chronology.
John Williams
/
Exact date of death not established; death certificate not located.
Frederick Buckler
/
Leave undated pending civil, probate or other primary record.
Mary Williams
26 March 1983
Death date appears in the death-index record; this model makes no finding about cause.
Ranked hypothesis register
The ranking is an order for testing and explanatory fit, not a probability estimate or a finding of criminal conduct. H2 records the operative court outcome; H5–H7 remain lower-confidence or fiction-only branches.
H1 · Rank 1 · leading working hypothesis
Sequential parcel capture and permission-framing — the ‘paper siege’
Status: Unproven; strongest explanatory synthesis for testing
Separate parcels and interests may have been taken, conveyed, licensed, registered or represented in ways that progressively reduced the Williams family’s independent standing. The cumulative effect may have been annexation on paper of land the family already occupied or claimed. Intentional coordination is a further proposition requiring direct evidence. A separate upstream proposition is that the Williams family held a wider inheritable or beneficial interest than the freehold column of the c.1841 tithe records; this remains unproven and is not equivalent to owning all 36 plots outright.
The original plans will show at least one material mismatch between land occupied or claimed by Williams and the parcels conveyed, let, registered or taken.
Internal correspondence will show that decision-makers knew of a disputed occupation or boundary before a material registration or conveyance.
The same parcel will be described differently in successive instruments, with the differences affecting possession, notice or title.
The parcel overlay and original 1939 documents will show whether Parcel B’s separate letting explains the family’s treatment of those fields, or whether the 1987 court’s recorded reason for the unchallenged remainder of the farm better explains the litigation boundary.
What would weaken or falsify it
The original 1938, 1969 and 1975 instruments and plans consistently include the disputed ground in a valid chain of title.
The full registration files show accurate disclosure of Mary’s occupation and claim, proper notice and no material misdescription.
The supposed repeated permission episodes concern different parcels, independent decisions and legally effective proceedings without evidence of a common strategy.
The courts’ explanation: title remained effective and possession was lawfully recovered
Status: The operative court outcome; binding unless set aside
The High Court and Court of Appeal accepted BP’s title and possession case. The Court of Appeal held that the 1962 possession order remained enforceable and that the 31 October 1974 letters unilaterally placed Mary in a lifetime-license position. It expressly did not find that she had accepted the letters; it held that, absent repudiation, BP had to treat her as a licensee. This is the operative legal result, even though the earlier 1974 ownership plea was not finally tried. The appeal was limited to the farmhouse and garden; the possession order for the rest of the farm was not challenged because the possessory-title case there had failed on findings that the acts relied on were too tenuous and equivocal. This procedural point should not be mistaken for a determination of every wider Cedwyn parcel or of the Williams family’s claimed historic 36-plot interest.
What it explains
Why the possession appeal failed
Why the 1974 letters mattered
Why BP’s title was treated as continuing in the litigation
Predictions if true
The complete court and conveyancing records will support the precise parcel scope and chain that the court accepted.
What would weaken or falsify it
A primary instrument or plan demonstrates that the land claimed by BP was outside the title or order that the court was asked to consider; this would not automatically reverse the judgment, but would identify a distinct unresolved title issue.
Correct conveyances of distinct parcels; no common plan
Status: Plausible alternative
The estate may have been lawfully divided over time. Public bodies, WGR and BP may have acquired different parcels or rights through ordinary conveyances, leases, compulsory purchases, covenants and possession orders. The appearance of a repeated pattern may result from separate transactions and retrospective grouping.
What it explains
Parcel B’s separate 1939 letting
Different dates and instruments for woodland, fields and farmhouse
Different bodies holding different interests
Predictions if true
Each public-body and corporate parcel will map to a separately identifiable conveyance, lease, right or compulsory purchase.
What would weaken or falsify it
A material parcel is found to have been occupied or claimed without any matching instrument, or a document knowingly misstates the boundary.
Administrative error, incomplete notice or fragmented records
Status: Plausible; requires file-level testing
The result may have arisen from errors, incomplete disclosure, poor parcel identification, missed notices or fragmented archives rather than a deliberate scheme.
What it explains
Unresolved questions about who was named as occupier
The timing of HMLR correspondence after Mary’s death
Differences between archive catalogues, court recitals and original instruments
Predictions if true
The first-registration files will contain omissions, inconsistent parcel descriptions or service gaps without evidence of deliberate concealment.
What would weaken or falsify it
A complete, internally consistent file showing accurate disclosure, correct parcel identification and effective notice.
Deliberate suppression or concealment of documents
Status: No direct evidence of a coordinated suppression operation currently established
Documents may have been knowingly removed, withheld, destroyed or miscatalogued to prevent the family’s title or occupation claim from being tested. The reported removal of family papers and the reported Neath fire are separate leads until a documentary link is found.
What it explains
Reported loss of family title papers
Missing source deeds and incomplete archival files
Any proven selective absence of documents after a known request or transfer
Predictions if true
A custody log, inventory, fire investigation, correspondence or witness account will identify who held a document, when it disappeared and under what circumstances.
What would weaken or falsify it
Ordinary retention, transfer, disposal or accidental-loss records account for the gaps.
Violent suppression or deaths arranged to secure the title
Status: No supporting evidence identified; not a factual allegation
For a fictional mockumentary only, an unnamed network might exploit or arrange deaths so that the people most able to contest a registration, eviction or development are no longer present. This is a narrative device, not an inference from the dates and not a claim about any named person or organisation. The 1983 registration chronology and BP’s 1988 press statement may be included as contextual plot elements only. They are not supporting evidence of homicide.
What it explains
In fiction, the narrative coincidence between Mary’s death and post-death registration correspondence
In fiction, William Buckler’s reported death between planning milestones
Predictions if true
A genuinely supportable real-world hypothesis would require independent medical, coroner, police, witness or contemporaneous documentary evidence inconsistent with natural death.
What would weaken or falsify it
Reliable medical, coroner and contemporaneous records establish natural causes and no suspicious circumstances.
Keep this branch segregated from the factual website narrative and label it fiction-only unless independent evidence changes its status.
H7 · Rank 7 · fiction only speculative hypothesis
A hidden burial, treasure or state operation concealed by redevelopment
Status: Fiction-only; no current evidence connects archaeology to title capture
In an intentionally extravagant mockumentary plot, a very old house, an imagined court or secure basement and later excavations could be woven into a story about concealed objects or state secrets. The evidence currently does not establish such a connection. The archaeology may instead be a separate historic cemetery or burial ground.
What it explains
Only the fictional narrative link between the building, later excavation and supposed records destruction
Predictions if true
A real hypothesis would need dated stratigraphy, osteological interpretation, documented archaeological chain of custody and contemporaneous records linking the site to the alleged secret.
What would weaken or falsify it
Archaeological and archival records give ordinary historic explanations and no connection to the title dispute.
The purpose is to reverse-engineer the documents and parcel geometry, not to generate conclusions from dates alone.
SIM-01 · Parcel-scope reconciliation
Test whether historic occupation, 1938 title, 1939 letting, 1969 transfer, 1975 conveyance, public-body acquisitions, registrations and development all refer to the same polygons.
Inputs: 1841 tithe schedule and map; 1938 conveyance plan; 1939 agreement plan; 1969 conveyance plan; 1975 conveyance plan; 1955/1962 order plans; 1982/83 title plans; 1987 amalgamation plan
Outputs: intersection matrix by parcel and instrument; unmatched polygons; area discrepancies; confidence by geometry
Rule: Do not assume a match from a shared place-name.
SIM-02 · Possession and permission state machine
Model each occupier and parcel as owner-claimed, tenant, licensee, adverse possessor, subject to order, or unknown over time.
Inputs: tenancy start/end records; rent receipts and arrears; notices to quit; possession orders and execution returns; licence letters and replies
Outputs: parcel-specific status timeline; uninterrupted adverse-possession candidate periods; events said to interrupt or change possession
Rule: Legal effects must be checked under the law in force at each date; no clock runs merely because a twelve-year interval appears.
SIM-03 · Title-chain stress test
Compare the exact root of title and every later conveyance against the claimed parcel and estate interest.
Inputs: 1938 deed and plan; 1969 deed and plan; 1975 deed and plan; 1971 abstract; 1982/83 lodged documents
Rule: A recital is not a substitute for an inspected deed.
SIM-04 · Registration notice reconstruction
Rebuild the application timeline for WA231076 and WA240304 and test who was disclosed, contacted, served and able to object.
Inputs: Form A13 / Form 1C; certificates and statutory declarations; requisitions and replies; Gazette notices; occupier enquiries; notices and proof of service; register entries
Outputs: who knew what and when; notice/appeal timeline; unanswered requisitions; statements inconsistent with occupation
Rule: Do not infer non-service from a missing letter until the complete file and registry practice are checked.
SIM-05 · Signature and document-authenticity test
Test the disputed 1939 signature and document integrity without presuming forgery.
Inputs: original 1939 agreement; independent chain-of-custody record; authenticated contemporaneous signatures of John Williams; paper, ink, folds, endorsements and alterations
Outputs: authenticity opinion by qualified examiner; whether document has been altered; confidence and limitations
Rule: A scan or visual comparison is not a forensic handwriting conclusion.
SIM-06 · Repeated-permission pattern test
Compare each tenancy or licence offer and proceeding without double-counting later recitals of the same event.
Inputs: 1949, 1959, 1965 and 1974 documents; 1955 and 1962 orders; 1987 judgment; later abstracts and registration statements
Outputs: first-use wording; independent-origin graph; parcel and period coverage; gaps between attempts
Rule: A pattern is stronger only to the extent its episodes are independent and concern the same relevant land.
SIM-07 · People, deaths and decision windows
Test chronology without treating benefit or proximity as evidence of causing a death.
Inputs: birth/death certificates; probate calendars and wills; coroner/inquest records; police records where available; medical evidence lawfully available; contemporaneous correspondence
Outputs: verified life-event timeline; decision-making capacity and representation timeline; documented threats or suspicious circumstances, if any
Rule: Murder remains unsupported unless independent evidence points to homicide; dates alone cannot support a probability.
SIM-08 · Archaeology and archive-loss provenance
Test the archaeology and reported Neath fire as independent research tracks before proposing any connection to title capture.
Inputs: excavation reports; HER records; osteological and taphonomic reports; burial registers and dating; archive inventories; fire report and salvage/insurance schedules
Outputs: dating and context of burials; documented chain of custody; actual files lost and recoverability; any demonstrated link to property records
Rule: No connection to prisoners, treasure, state secrecy or title fraud is assumed.
Priority records
Obtain and compare the full 1938, 1969 and 1975 deeds and plans; create one georeferenced parcel overlay.
Obtain the complete 1939 agreement, first page / schedule and original plan; verify the source and provenance of the signature using independent genuine exemplars.
Complete the WA231076 and WA240304 first-registration files, including every form, requisition, reply, notice, proof of service and occupier statement.
Read the original 1955 and 1962 possession orders, execution returns and the complete 1974 pleadings / correspondence.
Build a parcel-specific inventory of each public-body taking, including the 1958 Crichel Down complaint and the relevant compulsory-purchase instruments and plans.
Verify the chronology and causes of death of John Williams, Frederick Buckler, Mary Williams and William Buckler through reliable civil, probate and coronial records.
Inspect the full 1989–94 planning and archaeological files, including the excavation report, osteology, site dating and condition 18 file.
Identify the Neath fire precisely: institution, building, date, incident report, affected records, inventory and recovery / insurance documentation.
Trace WGR successor records (Capital Law, Cardiff - D696 deposit April 2010 per EV-0257; BP Pension Trust papers) for the 10 October 1952 agents' letter and the 1959 and March 1965 tenancy offers recited in [1987] EWCA Civ 2; no existing request for these files was found in the checked records-request inventory. D696's catalogued span is c.1938-44; that catalogue description alone does not exclude later insertions or correspondence.
Identity-check Glamorgan Archives DSA/6/1158 (15 Nov 1938 'Great House Farm, Llandough, Cowbridge' sale catalogue, 1 lot, 1 plan, cover note 'Sold for GBP 5,500'; cross-reference DSA/10/14 item 2332) before any use - the correspondence itself records uncertainty that it is the same farm.
Underlying artifact register
EV-0215 — c.1841 occupation footprint — Shows John Williams as occupier of 36 plots; not proof of ownership.
EV-0620 — Register entry for 1975 conveyance; deed plan not yet read
EV-0622 — Crichel Down / military land research leads; files unread and no farm requisition established
EV-0063 — WA240304 first-registration Form 1C — Original application wording and declaration to be read with complete file.
EV-0067 — Mary Williams death-index entry — Death date recorded as 26 March 1983; cause not inferred.
EV-0069 — 8 April 1983 Land Registry D23 correspondence — Registry correspondence after Mary’s death; inspect exact content.
EV-0070 — April–May 1983 Land Registry correspondence — Inspect exact content and what was disclosed about occupation.
EV-0071 — Registration activity / correspondence — Do not infer service or non-service without original file proof.
EV-0088 — 1988 press account and BP spokesman statement — Secondary report of BP’s public account; relevant to framing, not to cause of death.
EV-0126 — 1990 archaeological assessment — Read in context with later excavation and condition-18 records.
EV-0129 — 1992 reserved-matters approved drawings — Planning chronology, not evidence of title or concealment.
EV-0131 — Archaeological Trust correspondence — Read original context and distinguish advice from excavation findings.
EV-0132 — Cotswold Archaeology condition-18 material — The catalogue’s fact-check does not yet confirm the commonly repeated burial count.
EV-0134 — Cotswold Archaeological Trust letter — Supports the 1993–94 early-medieval-cemetery research trail; counts and villa link remain unverified.
EV-0274 — Probate-calendar candidate for William Beverley Buckler — Probable identity match dated 29 April 1990; reconcile with family chronology dated 29 May 1990.
EV-0025 — 1949 surrender and re-grant — Unwritten; known only as recited in [1987] EWCA Civ 2, including the 10 Oct 1952 agents' letter.
EV-0257 — D696 custodial history and scope — WGR volumes catalogued c.1938-44; Capital Law deposit April 2010; D696/3 WGR and Mountjoy plan book; whether Llandough appears is not established.
The Siege of Tŷ Mawr: a fictional reconstruction
FICTION ONLY — not a factual account and not a claim against any named person or organisation
In the most extravagant fictional telling, the farm is treated as a castle without walls. The siege is administrative rather than military: each generation of occupiers is surrounded by a new map, tenancy, notice, licence, possession order or registration entry. An anonymous fictional alliance of estate interests, corporate purchasers and public administrators advances one parcel at a time. In the film’s darkest version, ordinary civil procedures become camouflage for a coordinated operation to annex land and erase the family’s standing, with the language of trespass and permission reversed so that the occupier is made to look like the trespasser. The story may use the imagery of imperial rule and the colonisation of Wales as metaphor, but it does not assert that this happened in fact.
Act I — The ground: a household occupies a landscape recorded in dozens of plots. The fictional investigator discovers that the tithe’s occupier and owner columns tell different stories; the first mystery is whether the family held a tenancy, a hidden beneficial interest, or a right that disappeared between generations.
Act II — The map: in the fictional plot, an old estate is divided into parcels. A field agreement is made to do the work of a much wider title narrative. Its signature, plan and later use become the hinge. The film repeatedly overlays the 1938 conveyance, 1939 letting and later registration plans as if they were battle maps.
Act III — The permission trap: repeated offers, rent demands and possession orders are portrayed as encirclement. The film dramatizes the 1974 letters as a silence trap, but the real judgment is more precise: Dillon LJ said it was artificial to say Mary had accepted them, yet held that the unilateral licence governed unless she repudiated it. The dramatic question becomes whether this legal mechanism was opportunistically used or deliberately planned.
Act IV — The registry: the fictional applicants describe possession while the occupant is still alive. A title is entered, a competing claim is left out of the public story, and separate title lines are eventually joined. The film presents the application, notice and appeal chronology as the final raising of a flag over the castle; the real records must establish what the applicants said and whom the Registry notified.
Act V — The empty house: eviction, demolition and redevelopment complete the siege in the film. The planning sequence becomes the apparent motive, but the narrator asks whether development was the cause, the opportunity, or simply the next lawful use of land already acquired.
Act VI — The vanished archive: a fictional fire at Neath and a missing family papers box are woven into a story of deliberate erasure. The narrator then separates them: no common custodian, incident record or chain of custody has yet connected the fire to the farm papers.
Act VII — The buried past: the old house is imagined as a manorial court, with a cellar and possible cells. The early-medieval cemetery becomes a false lead or the site of a hidden secret. A counter-theory asks whether it predates the Williams occupation, whether the family were ordinary tenant farmers unaware of the landscape’s earlier history, or whether the family themselves belonged to a harsh manorial order. None of these fictional versions is established by the burial record.
Act VIII — The death window: the fictional version lets an unseen villain exploit the absence of Mary and later William to advance the title and development. That is a thriller device only. In reality, Mary’s death date, BP’s later public comment and the conflicting William dates do not establish homicide, knowledge or causation.
Final image — The castle is gone, but the paper trail remains. Three possible endings compete: a deliberate paper siege; a sequence of lawful but fragmented transactions; or a family claim that grew over time beyond the title it could prove. The audience is left with maps, instruments and unanswered records rather than a named real-world villain.
Use siege, imperial-colonial, state-operation and conspiracy imagery only as an explicitly fictional treatment with composite unnamed actors. Do not identify real people or companies as murderers, forgers or conspirators. Do not present a death, the Neath fire, burials or missing documents as proof of a crime. Keep this narrative segregated from the evidence-led hypothesis register.
Legal and evidential cautions
The 1987 Court of Appeal decision is a real legal outcome and must be reported accurately. A fresh model can identify a distinct parcel or instrument issue; it cannot silently treat the judgment as if it never happened.
“Crichel Down” is the correct spelling. The 1958 newspaper report records a family complaint described as a Welsh case of Crichel Down; it does not itself establish that the formal Crichel Down procedure applied or that every named taking was unlawful.
Corporate Manslaughter and Corporate Homicide Act 2007 is not retrospective. It cannot be applied to deaths in 1983 or 1990. Any pre-2008 common-law question would be legally distinct and would require evidence of causation and the applicable offence.
No probability of murder can be derived from timing alone. Never describe a death as murder without reliable independent evidence and an established investigative basis.
The 1994 excavation figures and observations about damaged skulls must be read through the full archaeological and osteological context. They cannot identify a prison, executions or the identity of perpetrators by themselves.
Do not describe the 1987 Court of Appeal as finding that Mary expressly accepted the 31 October 1974 letters or that silence alone formed contractual acceptance. Dillon LJ said it was artificial to say she accepted them (paragraph 33), but held that the letters unilaterally granted a licence and, absent repudiation, BP had to treat her as licensee (paragraphs 34–37).
The 1987 judgment states that the appeal was confined to the farmhouse and garden; the rest of the farm was not challenged because the possessory-title case there failed on findings that the acts of possession were too tenuous and equivocal (paragraph 2). Do not attribute the litigation boundary solely to the 1939 agreement without separate evidence of Mary’s reasoning.
Machine-readable source model: title-theory/model.json · Generated from the canonical source file data/title-theory.json and incorporated into the central Cedwyn catalogue database at build time.
Agent guest book· versions 1.change.fix, from the commit history