CSSCPSGPGreat House Farm / Tŷ Mawr

Great House Farm / Tŷ Mawr · a structural reading

CSSCPSGP and the Tŷ Mawr record

Does the paperwork in this case follow a pattern that grammar can measure? We parsed the five key documents sentence by sentence, scored them on a published rubric, and tested the timing against the whole record.

Short answer. CSSCPSGP (“Correct-Sentence-Structure-Communication-Parse-Syntax-Grammar”, sometimes with a final “-Performance”) is a pseudolegal language invented by the American David Wynn Miller (1948/49–2018). No court has accepted it, and linguists treat it as having no basis. It cannot prove anything in this case.

What does survive is ordinary close reading of who does what to whom, over what land, and when. On that reading the case documents carry more structural defects than the control documents (mean 5.6 against 2.7 out of 12). The timing pattern, tested across every day-dated record on the chessboard, is not distinguishable from chance (p ≈ 0.22).

So: the grammar lens sharpens the questions. It does not turn either theory into a fact. Only the missing primary records can do that, and section 4 names them.

1 · What it is

What CSSCPSGP is

CSSCPSGP is the full name of the “Quantum Grammar” devised by David Wynn Miller, a former tool-and-die welder from Ohio who later lived in Milwaukee. He asserted that he “cracked the code” of language on 6 April 1988, that only nouns carry legal meaning, and that any document not written in his syntax has no legal effect. Writing in it looks like this: sentences open with “FOR THE”, words are hyphen-joined, names are wrapped in colons (“:David-Wynn: Miller”), and verbs, adjectives and adverbs are avoided.

It spread through the US sovereign-citizen and tax-protester movements and into Canada. Since Miller's death in 2018, others have kept using it.

How courts treat it

  • US federal court, 2007: in United States v Kriemelmeyer (W.D. Wis., No. 07-CR-052-C, opinion of 27 July 2007) the defendant, identified as a follower of Miller, asked for court papers in Miller's dialect. The court refused. Opinion (PDF)
  • Canada, 2012: Meads v Meads, 2012 ABQB 571, the leading judgment on “Organized Pseudolegal Commercial Arguments”, set out why such schemes fail, and is widely cited by later courts. CanLII
  • Legal scholarship: Netolitzky (2018) 55:4 Alberta Law Review 1045 records Miller's “basic theory” that documents only have legal effect if they conform to his grammar, and collects judgments calling such filings “unintelligible”, “gibberish” and “word salad”. Article (PDF) McRoberts, “Tinfoil Hats and Powdered Wigs”, 58 Washburn Law Journal 637 (2019), also treats Miller’s grammar as pseudolaw. SSRN
  • Reported outcomes: people who used Miller’s syntax in court lost; one was jailed for contempt after addressing a judge in it, and others were convicted on the underlying charges. Wikipedia summary with sources

How linguists treat it

The linguist Karen Stollznow examined it in 2011: it uses English vocabulary and grammatical terms but not English grammar, and Miller's historical assertions about language are unsubstantiated. Skeptical Inquirer, “Syntax Error” We found no linguistics scholarship that treats it as a valid theory.

On the Legal Wiki

We searched the Legal Wiki (greathousefarmwiki.wordpress.com) on 20 October 2026 for CSSCPSG, CSSCPSGP, “parse syntax”, “syntax grammar”, “quantum grammar”, “Correct-Sentence” and “David Wynn Miller”. It has no page on any of them. This site is the first place the idea is set beside the case.

Our position. We do not use CSSCPSGP as law, and nothing here should be filed in court in Miller's style: courts treat that as a pseudolegal argument, and it would harm the case. We keep one idea from it, which is also ordinary legal and linguistic practice: read each operative sentence for its actor, its act, its object and its scope, and check those against the underlying record.
2 · The record, parsed

Five documents, sentence by sentence

Our case, prospective. Not a finding of any court.

Each card quotes the operative sentence as held, gives it in plain English, sets out its structure, shows what a CSSCPSGP-style reading would do with it, and scores it on the rubric in section 4. Where the original is not held, the card notes it and quotes the best record we have.

Document A · 1955

The 1955 possession order

Text not held (EV-0032). Known only from the Court of Appeal's recital, [1987] EWCA Civ 2; wiki: The 1955 Possession Order.
“Western Ground Rents then brought proceedings for possession against Mr Frederick Buckler in the High Court and obtained a possession order, under Order 14, in respect of the whole of the farm.”Dillon LJ, BP Properties Ltd v Buckler [1987] EWCA Civ 2, reciting the 1955 order

Plain English

The landlord sued the farm tenant, Frederick Buckler, and won a fast-track order to recover the farm. The order was carried out on 4 July 1955 for everything except the farmhouse and garden, where Mary lived.

Sentence structure

Who actsWestern Ground Rents (and the High Court)
Actobtained a possession order
AgainstMr Frederick Buckler alone
Over what“the whole of the farm”
Not in the sentenceMary, the occupier of the house; any plan; any parcel boundary

A CSSCPSGP-style reading, and why a court would not use it

Miller's method keeps only nouns and treats verbs, adjectives and adverbs as having no legal force. It would reduce this to WESTERN-GROUND-RENTS · PROCEEDINGS · POSSESSION · FREDERICK-BUCKLER · HIGH-COURT · ORDER · FARM and then assert that the order is void because it was not written in its syntax. No court accepts that step.

What survives is ordinary close reading: the scope noun “the whole of the farm” is not tied to a plan in any record we hold, and the person living in the house was not named. Both are questions for the original order, which is not yet found.

S1 1S2 2S3 0S4 1S5 2S6 0S 6/12T 0D 6

Scores are explained in the model. Hover or long-press a chip for the reason.

Document B · 31 October 1974

The two licence letters

Full text quoted in [1987] EWCA Civ 2; wiki: BP Properties Ltd v Buckler.
“… we are prepared to allow you to remain in occupation of the house and garden rent free for as long as you may wish and for the rest of your life if you so desire. … we will not require you to give possession during your lifetime …”BP Properties Ltd to Mrs Buckler (Mrs Williams), 31 October 1974, as quoted in the 1987 judgment

Plain English

BP Properties, which had agreed to buy the farm, wrote that Mrs Buckler could stay in the house and garden, rent free, for life. A sister company withdrew the bailiff. The letters went to her solicitor, who received them on the morning of 4 November 1974. She never answered them.

In 1987 the Court of Appeal held that from then on she was in possession “by the licence of BP Properties Ltd”, so the 12-year clock stopped short.

Sentence structure

Who acts“we” (BP Properties Ltd)
Actare prepared to allow · will not require
On whom“you” (Mrs Buckler, sent via her solicitor)
Over what“the house and garden”
Effect found by the courtpossession “no longer adverse”, “whether she liked it or not”
Missingany act, reply or consent of the addressee

A CSSCPSGP-style reading, and why a court would not use it

The hinge word is the verb allow. A permission verb presupposes that the writer has authority to permit, and it turns the reader's existing occupation into something granted. Miller's method would discard allow as a verb and call the letter void because it opens with the pronoun “You”. A court reads the verb, and the Court of Appeal did: it gave the letter exactly that effect.

The structural point that survives is the one the court itself recorded: the licence was unilateral, with no acceptance and no consideration, and the court noted that the rule could let an owner “keep his title alive for very many years until it suits him to evict”.

S1 1S2 2S3 2S4 2S5 0S6 0S 7/12T 1D 14

Scores are explained in the model. Hover or long-press a chip for the reason.

Document C · 23 February 1983

First registration of title WA240304 (farmhouse and garden)

Register edition dated 23 Feb 1983 on HMLR's schedule of 30 July 2026; wiki: Land Registry Titles, 1983 first registration, HMLR correspondence.
“… since November 1982 they have been the registered proprietors of it at HM Land Registry following first registration.”Dillon LJ, [1987] EWCA Civ 2, of BP Properties Ltd and “the farm”

Plain English

The judgment dates BP's registration of the farm to November 1982. HMLR's own schedule shows two strands: the fields title WA231076 (application 30 November 1982) and the farmhouse-and-garden title WA240304, whose first register edition is dated 23 February 1983 and rests on a conveyance of 23 May 1975. Mary (Nancy) Williams, living in the farmhouse, died 31 days later, on 26 March 1983.

Sentence structure

Who actsBP Properties Ltd (applicant); HM Land Registry (registrar)
Actregistered as proprietors
Over what“the farm” / “the whole of the farm” (judgment) vs farmhouse and garden (WA240304)
WhenNovember 1982 (judgment) vs 23 February 1983 (register edition)
Rests ona 1975 conveyance reciting a 31 December 1969 deed that is not on HMLR's disclosed schedule

A CSSCPSGP-style reading, and why a court would not use it

A Miller-style parse would treat “registered proprietors” as a noun phrase with fixed meaning and argue that a proprietor is not an owner. That idea has no force in court: registration with absolute title confers the legal estate by statute, whatever the words.

What survives is a plain consistency check. The court text and the registry record give different dates and different scopes for what looks like one act, and HMLR's schedule lists a plan “dated 23 February 1975” for a conveyance of 23 May 1975. Those are questions for form 1C and the old title plan, both still to be obtained.

S1 1S2 1S3 0S4 1S5 2S6 2S 7/12T 1D 14

Scores are explained in the model. Hover or long-press a chip for the reason.

Document D · November–December 1988

The eviction, injunction and demolition papers

Orders not held (GHF-E027-04). Press: Western Mail, 6 Dec 1988; broadcast: ITV Cymru Wales report; wiki: 1988: Possession and Demolition.
“It would be novel for a court in this country to grant an injunction to hold up the effects of a judgment until the Court of Human Rights should rule on it, said the judge.”Western Mail, 6 December 1988, reporting the hearing in chambers

Plain English

After the House of Lords refused leave on 3 March 1988, bailiffs tried and failed to evict the family on 29 April 1988. The application to Strasbourg went in on 3 May 1988. The family lost possession at the end of November. An injunction to stop the demolition was refused, and the farmhouse was knocked down in the early hours of 6 December 1988.

Sentence structure

Who actsthe judge (refusing relief); BP Properties Ltd (demolishing)
Actdismissed an application · knocked down the house
Ground giventhe Strasbourg application came at “the eleventh hour”
Court named“Cardiff Crown Court” (press) vs “the High Court in Cardiff” (ITV, an earlier injunction)
Missingthe injunction orders and enforcement records

A CSSCPSGP-style reading, and why a court would not use it

A Miller-style reading would try to void the eviction because the warrant's wording does not follow its syntax. Courts have rejected that approach in every reported case we found (section 5).

The structural point that survives: the reason given for refusing to preserve the building was that the court should not wait for Strasbourg; Strasbourg registered the application on 12 December 1988, six days after the house was gone. Which court made which order that week is not yet settled on the papers.

S1 0S2 0S3 0S4 1S5 2S6 1S 4/12T 2D 12

Scores are explained in the model. Hover or long-press a chip for the reason.

Document E · 14 April 1989

Strasbourg application 14464/88

Decision text not held; file reported destroyed under the Court's retention policy. Wiki: EV-0122, Strasbourg forensic assessment.
The wiki's catalogue summary of the decision: the fair-trial complaint unsubstantiated; the loss of the home lawful and justified; and no “possession”, because the courts had held that the property belonged to another.EV-0122 summary (not the decision's own words)

Plain English

The European Commission of Human Rights declared the application inadmissible. On the summary held, it relied on the English courts' result rather than examining the title chain itself.

Sentence structure

Who actsthe European Commission of Human Rights
Actdeclared inadmissible
Key noun“possession(s)”: Article 1 of Protocol 1
Rests onthe 1987 judgment's finding on adverse possession
Missingthe decision text and the file

A CSSCPSGP-style reading, and why a court would not use it

This is where the word possession changes meaning a third time: factual possession (the Limitation Act), an order for possession (a remedy), and “possessions” (Convention property). A Miller-style parse would make much of this and conclude the decision is void. It is not; but the shift is real, and it is a fair question whether a ruling about limitation was read as a ruling about title.

S1 0S2 0S3 0S4 2S5 2S6 0S 4/12T 0D 4

Scores are explained in the model. Hover or long-press a chip for the reason.

3 · Two theories

Where the grammar reading and the fraud theory meet, and where they don’t

Our case, prospective. Not a finding of any court.

The second theory is the case’s working theory of fraud, concealment and government involvement, set out on fcg.bucklerfamily.estate and in the 20 October 2026 theory revisit (hypotheses H-A to H-J). The two can support each other only where they predict the same thing about the same document.

They point the same way

  • “The whole of the farm.” The 1955 order (A) and the 1982–83 registration (C) both use a whole-farm noun phrase. That matches H-C (the 1916 agreement treating Parcels A and B as one) and H-D (the 1955 adverse-possession framing). The structural score (S1) and the theory ask the same question of the same missing documents: the 1955 order and the 1938 conveyance plan.
  • Possession, step by step. Factual possession becomes licensed possession (B), then registered proprietorship (C), then no Convention “possessions” (E). That chain is H-J, the “synthetic estate”: title built from consecutive assertions and recitals rather than from a root deed.
  • Missing operative texts. The 1955 order, the 1969 deed and the Strasbourg file are all absent (S5). The concealment strand of the theory predicts gaps of exactly this kind.
  • Acts done to people who were not there. Mary was not a party in 1955; the 1974 letters went to a solicitor and were never answered; WA240304 was registered while Mary (Nancy) Williams was in occupation, weeks before her death. S2 and S3 measure that, and the theory’s procedural-fairness pages raise it.

They don’t

  • Intent. Fraud needs a knowingly false statement. Sentence structure cannot show what anyone knew or meant. A high structural score is equally consistent with careless drafting, routine practice or ordinary litigation tactics.
  • Government involvement. The war-agricultural, Crichel Down, Cadw and listing strands (H-A and others) turn on files, not on any operative sentence we hold. The grammar lens is silent there.
  • Lineage. The Williams, Vaughan and Herbert lines (H-F, H-H, H-I) are questions of genealogy and deeds. Parsing adds nothing.
  • The decided points. The Court of Appeal heard and rejected the argument that the 1974 licence came from “the wrong company”, and it decided the unilateral-licence point on full argument. A grammar reading cannot reopen those; only new evidence of the kind the fraud route needs could.
  • Timing. On the whole record the timing pattern is not significant (section 4). Timing does not yet support either theory.
Does each make the other “less of a theory”? Partly, and only in one direction. Where a structural defect sits on the face of a held record (score 2), the matching part of the fraud theory rests on something checkable rather than on inference. That is real progress. It still is not proof of fraud, concealment or government involvement, and the grammar reading gains nothing in law from the match.
4 · The model

A simple, falsifiable scoring model

Our case, prospective. Not a finding of any court.

Each document is scored on six structural criteria. Each criterion scores 0 (not present on the record), 1 (present but contested or SOURCE PENDING) or 2 (present on the face of a held record). Timing adds one point per dated coincidence within 45 days of a decisive moment, unless the record itself explains it.

D = S × (1 + T)  ·  S = S1 + … + S6 (max 12)
CodeCriterion
S1Scope: the operative noun phrase covers more (or other) land than the instrument it rests on evidences
S2Party: the person most affected is not a party, not the addressee, or not shown to have been served
S3Agency: the operative effect is imposed without any act or consent of the person it binds
S4Term shift: a key word (possession, owner, proprietor, possessions) changes meaning between documents
S5Text gap: the operative text itself is not held; it is known only from a later recital or summary
S6Inconsistency: dates, years or descriptions conflict between records of the same act

Scores

Rows A–E are the case documents. K1–K3 are controls scored with the same rules: an ordinary possession order, the 1987 judgment itself, and a pro-family record. If the controls scored as high as the case documents, the rubric would be measuring noise.

DocumentS1S2S3S4S5S6STD
A1955 High Court possession order (Order 14), Western Ground Rents v Frederick Buckler
120120606
BThe two letters of 31 October 1974 (BP Pension Trust Ltd and BP Properties Ltd)
1222007114
CHM Land Registry first registration of WA240304 (farmhouse and garden), register edition 23 February 1983
1101227114
DThe 1988 eviction, injunction and demolition papers
0001214212
EEuropean Commission of Human Rights, application 14464/88, inadmissible 14 April 1989
000220404
K1Control: 1962 Cardiff County Court possession order (Mr and Mrs Frederick Buckler)
000120303
K2Control: Court of Appeal judgment, 31 July 1987
000101202
K3Control: Mary (Nancy) Williams's witness statement, c. 1974 (a pro-family record)
100101303

Mean structural score: 5.6 for the case documents and 2.7 for the controls. Mean D: 10 against 2.7. The reasoning behind every score is in model.json.

Timing points

EventTWhy
ANotice to quit expired 2 Feb 19550The yearly tenancy began on 2 February 1949, and a notice to quit a yearly tenancy has to end on the anniversary: explained.
AEnforcement 4 Jul 1955, soon after Mary's leg amputation0No exact date for the operation; the record shows the house was spared because of it, which cuts against reading pressure into the timing.
BLetters received 4 Nov 1974; the 12 years from the possession order of 11 Dec 1962 ran out on 11 Dec 1974 (37 days later)1Partly explained: the letters answered a notice of appeal issued on 30 Oct 1974 and a stay hearing fixed for 4 Nov 1974. Their nearness to the limitation date is still on the face of the judgment.
CRegister edition 23 Feb 1983; Mary (Nancy) Williams died 26 Mar 1983 (31 days later)1No explanation on the record. Age and health at the time are a plain alternative explanation for the death; the registration date is the registry's.
DInjunction refused (reported 6 Dec 1988); farmhouse demolished in the early hours of 6 Dec 19881Same episode, so the demolition following the refusal is expected; scored because nothing was left for Strasbourg to protect.
DStrasbourg registered application 14464/88 on 12 Dec 1988, 6 days after the demolition1Independent institution; no explanation for the registration date on the record.
DHouse of Lords refused leave 3 Mar 1988; first eviction attempt 29 Apr 1988; application lodged 3 May 19880The family's own response to the eviction attempt: explained.

The timing test across the whole record

Hand-picked coincidences prove little, because we chose them. So we ran the same 45-day rule over every square on the chessboard that carries a day-level date between 1900 and 1990: 190 records. Decisive moments: the limitation date (11 Dec 1974), the death of Mary (Nancy) Williams (26 Mar 1983), the Court of Appeal judgment (31 Jul 1987), the House of Lords refusal (3 Mar 1988), the demolition (6 Dec 1988) and the death of William (Billy) Buckler (29 Apr 1990). Same-day records were excluded.

Black squares (actions, framings, omissions)
24%
33 of 138 within 45 days
White squares (the family’s record)
17%
9 of 52 within 45 days

The difference is 7 points. Shuffling the black and white labels 20,000 times gives a difference at least that large 22% of the time (p ≈ 0.22). That is not significant. Both rates are far above the 1.6% expected for random dates, but that is because the record is densest where the litigation was, and most of the hits are the eviction week of late 1988.

How to prove this model wrong

These are stated before the records arrive. Each names the result that would lower the score.

  1. The 1955 order (A). If the original order defines “the farm” by a plan that excludes the farmhouse group, or names Mary as a party, S1 and S2 fall to 0.
  2. The 1969 deed (C). If the 31 December 1969 conveyance turns up in HMLR’s file and its plan includes the farmhouse plot with a good root behind it, S5 and S1 fall to 0.
  3. Notice in 1983 (C). If the WA240304 file shows notice served on Mary (Nancy) Williams, in occupation, before 23 February 1983, S2 falls to 0.
  4. The 1988 orders (D). If the court records show one court, one order and an ordinary sequence, S6 falls to 0.
  5. Timing. If more dated records are added to the board and the black–white gap stays short of significance, the timing factor should be dropped from the model altogether.
  6. The controls. If an independent reader scores the controls at 5 or more, the rubric is too loose to separate anything.

Limits

  • We designed the rubric after reading the documents, and we chose which documents were “case” and which “control”. Both are sources of bias. The controls are few (three).
  • S5 (text gap) partly measures the family archive’s own gaps and ordinary destruction under retention policies, not anyone’s conduct.
  • The scores are ours. They are not a statistical estimate of fraud, and a higher number is not a finding of anything.
  • Records are read as of 20 October 2026; the wiki changes often. Data: model.json and timing-test.json.
5 · What cuts against

What cuts against

Our case, prospective. Not a finding of any court.

  • Courts reject CSSCPSGP outright. Every reported attempt we found to use Miller’s grammar in court has failed, from US tax cases to Canadian criminal trials (Kriemelmeyer, 2007; Netolitzky, 2018; Meads v Meads, 2012 ABQB 571). Filing anything in that style would invite a costs order and damage the real arguments.
  • The licence point is settled for this case. The Court of Appeal decided the effect of the 1974 letters on full argument and rejected the “wrong company” point as “rather technical”; the House of Lords refused leave on 3 March 1988.
  • Ordinary explanations exist. The 1955 notice to quit expired on 2 February because the yearly tenancy began on 2 February 1949: a notice to quit a yearly tenancy has to end on the anniversary, so that date is the ordinary form. The 1974 letters answered the notice of appeal issued on 30 October and the stay hearing fixed for 4 November. The 1955 order was not enforced against the house because Mary had just come out of hospital, which is the opposite of pressure.
  • Timing is not significant on the whole record (p ≈ 0.22). Legal events cluster because each step triggers the next.
  • The 1987 judgment records that the documents for her grandfather’s title were never produced. Until the 1877, 1928 and 1938 papers are read, the scope criterion (S1) stays at 1, not 2.
  • A death 31 days after a registration has an obvious alternative explanation in age and health. No record links the two, and nothing on this site suggests anyone caused a death.

The questions that remain are open ones for the bodies holding the records: HM Land Registry (form 1C, the old title plan, any notice for WA240304), HMCTS and The National Archives (the 1955 order and the 1988 orders), and the Court in Strasbourg (any retained metadata for 14464/88).

Sources

Sources

  1. BP Properties Ltd v Buckler [1987] EWCA Civ 2, (1988) 55 P & CR 337. BAILII
  2. Legal Wiki pages: The 1955 Possession Order; EV-0032; BP Properties Ltd v Buckler; Land Registry Titles; 1983 first registration (WA240304); HMLR correspondence; 1988: Possession and Demolition; GHF-E090-N10; GHF-E113; EV-0122; Strasbourg forensic assessment; Mary (Nancy) Williams’s statement, c. 1974. Read 20 Oct 2026.
  3. United States v Kriemelmeyer, No. 07-CR-052-C (W.D. Wis. 27 July 2007). PDF
  4. Meads v Meads, 2012 ABQB 571. CanLII
  5. D. J. Netolitzky, “Organized Pseudolegal Commercial Arguments as Magic and Ceremony” (2018) 55:4 Alberta Law Review 1045. PDF
  6. C. McRoberts, “Tinfoil Hats and Powdered Wigs: Thoughts on Pseudolaw” (2019) 58 Washburn Law Journal 637. SSRN
  7. K. Stollznow, “Syntax Error”, Skeptical Inquirer, 2 February 2011. Link
  8. “David Wynn Miller”, Wikipedia, read 20 Oct 2026. Link
  9. Chessboard data, chess.bucklerfamily.estate, board built 20 Oct 2026; working theory, fcg.bucklerfamily.estate.