Treehouse — design & build · Adam W. Stauffer

Studies answer “should we?” They are illustrative, they may show options that were never adopted, and they must never be measured from. Each one is frozen at the datum it was drawn against — it is a dated record of what was known when a decision was made, so it is not refreshed when the design moves. For what is actually being built, see the Drawings tab.

✅ Decided — the options of record

Porch: Option E — the larger, full-width porch; the ladder lands in the NW quadrant and you step north into it. Ruled 2026-08-09 · open the frozen study
Chase-east plexi wall: P · PLINTH — one OPTIX 36×72 sheet upright, equal 8¼-in margins left/right/top, solid plinth below, sill 27.4; no seam; 98.6 % of the trunk in view. Ruled 2026-08-12 (gates: the acrylic datasheet + the open-shaft envelope stand) · open the frozen study
Blocking: the half-drop lattice — rows every 16 in per bay, alternate bays dropped 8 (odd 16/32/48, even 24/40); adjacent bays never share a height. Ruled R17, 2026-08-13 · open the frozen study
Wall height: 5'-6" FINAL — the default won over the 6-ft option; ledger closed balanced. Ruled R14, 2026-08-13 · open the frozen study
East wall: one window, the E-S1b envelope — RO 40×24, head 54, sill 30; found-window range 32–40×24; exit window centered (R13). Ruled R18/R19, 2026-08-13 · open the frozen study
South wall: SE-3 + G-A — the 27×32 egress hard east (the second means of egress of record; the siting deleted the ridge-post PE question) + a 33×26 fixed pane west of the ridge post, zero Σb. Ruled R18, 2026-08-13 · open the frozen study
Door-wall extension: Option B — sta 43.5, X 42→66.5, full height, sheathed, counted as a candidate; new solid 4×4 end post at X 42. Ruled R36, 2026-08-15, pending PE · open the frozen study
Plate corners: half-lap, lead — 2-in tongues pinned over the solid rough 4×4 corner post (R30); butt + HTP37Z the alternate; PE confirms both. Ruled R28, 2026-08-14 · open the frozen study
The plate ring — six posts, four 12-ft rough 4×4 sticks, two-segment gable plates side-framing the ridge posts; posts center on the square-frame nailers (R34). Ruled R32, 2026-08-14; the cathedral (R35c) later deleted the two station-16 segments · open the frozen study
North triangles — station 16: full cathedral (R35c); station 43.5: SHEATHED with a closable screened vent (R38). Ruled 2026-08-15/16 · open the frozen study
Post caps & bases — AC4RZ pairs at the mids, EPC4RZ at the gables, field-sprayed black; MSTA12Z strap pairs at the bases and shim joints. Ruled R35, 2026-08-15, pending PE · open the frozen study
The cathedral porch — open deck-to-rafters north of the door wall; ridge-line headroom 8'-11⅝". Ruled R35c, 2026-08-15 (datum v11, pending PE) · open the frozen study

The SOUTH gable, framed — R41's arrangement, drawn

⏸ ADOPTED IN PRINCIPLE (R41, 2026-08-16) — the owner is reviewing this study. Gable studs stacked over wall studs at 16 o.c., the rough 4×4 ring plate as the visual band, sheathed — ⭐ the aesthetic IS the code condition (stacking makes R602.3.2's ≤ 1-in offset zero). ⭐⭐ THE FIND: the triangle the framing makes is not register G-1's — its top boundary is the roof SHEATHING plane, a constant 5.0–7.3 in lift — so the RULED G-A window gains 12 % of glass, all in height, free (872 → 975 in²) and the ruling survives its own correction. Every detail above the plate stays PROPOSED (P1/P3/P10, PE).
⛔ FROZEN at datum v11 · 2026-08-16 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V11 (2026-08-15) — CONFIRMS AN ARRANGEMENT, ADOPTS NO DETAIL, SIZES NO HEADER, SELECTS NO PART — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. R41 ADOPTS THE ARRANGEMENT IN PRINCIPLE; EVERY DETAIL ABOVE THE PLATE IS PROPOSED (MEMO P1/P3) AND PE-GATED.

The south gable, framed

R41 adopted an arrangement in one sentence and asked to see it. This is that sentence drawn — wall studs at 16 in o.c. below, the rough true 4×4 ring plate as a horizontal band, gable studs standing on it on the same centers, bevel-cut to a raked cap at the sheathing plane. Drawing it turned up four things the sentence did not contain, one of which moves a figure the register calls decided.

THE BRIEF, VERBATIM (owner, R41, 2026-08-16):
“dimensional gable studs stacked above wall studs broken visually horizontally by 4x rough top plate. 16 oc same as wall studs so gable and wall studs look ‘continuous’; disappear into top plate but then reappear on journey up to the ceiling. sheathed.”

⭐ The aesthetic IS the code condition. The 2021 IRC R602.3.2 single-top-plate exception wants the framing above a single plate to sit over the studs below within 1 in. Stacking makes that offset zero. The look the owner described and the compliance route are the same discipline — which is why one drawing answers both, and why the arrangement is worth confirming rather than merely permitting.

PROVENANCE — what this page is, and is not

class: STUDY. Illustrative. It answers “is this the arrangement?”, not “how do we build it?”
datum: FROZEN at DATUM v11 (2026-08-15). Every model figure was derived from geometry.py at generation and is now a dated snapshot. ⛔ Do not re-run this generator on a later datum — write a new study.
status: R41 adopts the ARRANGEMENT in principle. Every detail above the plate — the raked cap, its orientation, the corner, the connectors, the header — is PROPOSED (memo P1/P3, section 1.4) and rides the PE gate. Nothing here is scheduled.
derived here: The stud FACE (1.5 in) and the 16-in module: the datum carries WALL_STUD_THK 3.5 (the wall depth) and no constant for either, because Year-2 walls are prose-only. Sources: registers 1.1a-rev and 1.1. NOMINAL, FIELD VERIFY. Everything else on the page derives from datum constants.
not this wall: ⛔ The triangle R38 has just sheathed and vented is at station 43.5, the door wall. This page is the south gable line, station 138 — the far end of the building.

① The elevation — the owner's line, drawn

The stud line runs from the sole plate to the plate underside at 5'-2", disappears behind 4" of rough true 4×4, and reappears on the same centers above it. The two gable-line plate segments (5'-6 1/2" + 3'-1 1/2") side-frame the continuous S-gable ridge post (J14); the corners over the R30 posts are J13. Neither joint is specified — both are PE TO CONFIRM.

⭐⭐ THE TRIANGLE THE FRAMING MAKES IS NOT THE TRIANGLE THE REGISTER MEASURES. Register G-1 (DECIDED 2026-08-10) rules gable heights off the slope-defining triangle — drawn dashed above — and its stated reason is that it closes on the plate at both wall lines at exactly 0.0000. The framed triangle does not close there at all: its top boundary is the roof sheathing plane, which stands 5.03 in above the plate at the SW corner and 7.27 in at the SE. Two different surfaces have been carrying one name.

The gap is not an error in either place — it is the eave seats plus the rafter's own plumb depth. West: the soffit plane passes through the plate top 2 1/2" inboard of the wall line and the rafter is 7.003 in deep measured plumb. East: 1 1/2" and 9.316 in on the steeper plane, which is why that corner stands higher. NEEDS_CONTEXT — owner. G-1's own words say “gable heights” without saying which question they answer, and for anything framed inside this triangle the framed surface is the binding one.

⭐ AND THE DIFFERENCE IS A CONSTANT, WHICH IS WHAT MAKES IT SAFE TO ABSORB. Both surfaces are planes at the same pitch, so the offset does not vary along the rake: on the west plane the framed sheathing plane runs a flat 5.033 in above G-1's line everywhere, and the buildable surface — under a raked cap laid flat — a flat 3.123 in; on the east plane, 7.265 and 4.725 in. The correction is a uniform lift, not a change of shape, which is exactly why every width sized off G-1 keeps its proportions and simply gains head-room — and why the ruled window's own width survives below. (The generator asserts the offset is constant at five stations; if that ever fails this paragraph is wrong, not stale.)

② The stack, in section

③ The inside face

The ruled window, and what the correction does to it

⛔ THE 1.2c/1.2d PAIR IS SUPERSEDED AND IS NOT DRAWN AS A CANDIDATE. R18 (owner, 2026-08-13) ruled the south above-loft opening a single fixed pane west of the ridge post — G-A, 33.25 × 26.21 — and the same-day EAST reopening left the SOUTH half standing confirmed. The pair's arithmetic (18.75 × 25.63 each, 961 in² together) stays the frozen studies' record; it is quoted here as history and nowhere as an option. The framing is drawn to accommodate the ruled opening.

the same 33.25-in-wide opening, measured against…	width	height	area	basis
G-1's construction (what G-A was RULED on)	33.25 in	26.21 in	872 in²	register G-1 + 1.2d — the slope-defining triangle
the FRAMED triangle, cap FLAT	33.25 in	29.33 in	975 in²	DERIVED HERE — sheathing plane less a flat 2x4 cap
the FRAMED triangle, cap ON EDGE	33.25 in	26.79 in	891 in²	DERIVED HERE — sheathing plane less a 2x4 on edge
area peak on the framed triangle, cap FLAT	35.23 in	27.77 in	978 in²	DERIVED HERE — not a proposal
area peak on the framed triangle, cap ON EDGE	33.62 in	26.50 in	891 in²	DERIVED HERE — not a proposal

⭐⭐ THE RULING SURVIVES ITS OWN CORRECTION — AND GAINS 12 % OF GLASS, ALL IN HEIGHT, FREE. On the framed triangle G-A's own ruled width is worth 975 in² against the 872 in² it was ruled on — and 33.25 in sits within 0.3 % of the real area optimum with the cap flat and 0.0 % with it on edge. The curve is flat at its top in both readings, which is the useful fact: the owner's width was chosen for looks and it is right for area too. ⛔ Nothing here re-rules G-A — it reports that the ruled opening has more head-room available than the arithmetic behind it assumed.

⚠ AND THE CAP NOBODY HAS SPECIFIED IS WORTH 85 IN² OF IT. Memo P1 says “a 2×4 raked top member set to the sheathing plane.” Flat or on edge is stated by no source, and it moves: the buildable glass by 85 in², the optimum width from 35.23 to 33.62 in, and the whole corner condition — at the SW corner the cap dies 3.12 in above the plate laid flat and 0.58 in on edge, which is to say it effectively lands on the plate. At the SE corner the same pair is 4.72 / 1.34 in. NEEDS_CONTEXT — PE, with P10. Nothing is invented: both readings are drawn, both are priced, and neither is chosen.

What this page raises

⭐⭐ G-1's construction and the framed triangle are different surfaces (5.03 / 7.27 in at the corners). Everything sized in this triangle off G-1 has that much more room than it was sized against. NEEDS_CONTEXT — owner; the register row is DECIDED and this page does not re-rule it.
⚠ The raked cap's orientation is unstated and expensive (85 in² of glass, and the entire corner condition). NEEDS_CONTEXT — PE, with P10, which is the same detail seen from the diaphragm side: the cap is the roof sheathing's boundary member at this edge.
⛔ G-A's east jamb IS the loaded S-gable ridge post (271.6 lb, continuous floor to ridge soffit, seated on J12 — PE TO CONFIRM). R18 chose SE-3 hard-east below the loft precisely to delete the ridge-post/king-stud question; above the loft, G-A re-creates it. PE.
⚠ Build-guide 2.11 is one datum generation behind. Its table prices the E & W plate as a doubled 2×4 reading WHITE, and its section 2 invents a 0.50-in facing board to carry the timber band round a corner it says the band dies at. R5 made that plate a single rough TRUE 4×4 — TIMBER by 2.11's own rule — and R32 put the same member on this gable line, so the band the owner is asking to see is already structural and the facing board is solving a problem the datum dissolved. The band does still stop — at the NORTH, because R35c deleted the station-16 segments. Domain C.
⚠ The stud layout ORIGIN is stated by no source. Drawn from the west wall line on the sheathing module, which puts the balance bay (11 1/4") at the east end. Which end carries it decides whether G-A's jambs land on the module, and it is an owner/field call, not a drafting one.
⬇ Does the SOUTH triangle get its own operable vent? R38 gave the station-43.5 triangle a closable screened vent for summer airflow. This gable is the other end of the same roof cavity, and a single vent is half a path. OPEN — owner question, not asked here and not answered here; any vent is an operable opening and lands in the enclosure inventory (ASCE 7-16 §26.2), which is a PE line.
⛔ No header depth and no rough-versus-net-clear allowance exist on file for this opening — the frozen south-wall-openings study raised both and neither has been answered. Nothing is sized here.

The verdict on the arrangement itself: it draws cleanly, it is the conventional gable-endwall detail for the condition (full-height gable studs on a raked cap — AWC Guide to Wood Construction in High Wind Areas §4.2, Fig. 10 + Table 5; the hinge-at-the-plate-line failure FEMA P-804 §4.2 exists to retrofit is answered here by continuity at construction), its 16-in module clears the continuous ridge post at every station, and it delivers the horizontal timber band the interior finish document was building a workaround to fake. Nothing found argues against it. The seven items above are what it costs to draw it honestly, and the owner rules every one of them.

STUDY — illustrative, frozen at DATUM v11 (2026-08-15), generated 2026-08-16 from geometry.py. PLANNING — NOT FOR CONSTRUCTION. R41 adopts the arrangement in principle; every detail above the plate is PROPOSED and the owner rules. ⛔ Do not re-run this generator on a later datum — write a new study.

One larger east window — the ruling page

⏸ ENVELOPE RULED (R19, 2026-08-13) — the unit is still being selected. R18-EAST reopened the pair the same day it was ruled; R19 then fixed the one-window envelope this study derived: RO 40×24, head 54, sill 30 — the calcs bind the 40-in max (worst mechanism 0.689 with R16's nailing) with an accepted found-window range of 32–40×24 (salvage welcome; 32 is the footnote-immune floor). The finalists, sills and rafter-tail sightlines here are the selection's evidence. REV 2026-08-13: finalists + sills + sightlines added at owner request.
⛔ FROZEN at datum v9 · 2026-08-13 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — NO WIDTH IS SELECTED, NO HEIGHT IS ADOPTED, NO SILL IS SET, NO HEADER IS SPECIFIED, NO MECHANISM IS ADOPTED — GENERATED FROM geometry.py AND A RECALCULATED calcs.xlsx, NOT HAND-DRAWN

One larger east window — instead of two

On 2026-08-13 R18 ruled a four-window program, and the same day the owner reopened its east half: “i like the idea of one larger e window rather than two on east”. So the ruled pair — the centered 24 × 24 squeeze window plus a contiguous 16 in high band, 40 in of opening — is now the reference, and this page asks what one hole can do instead in the 94.5 in of wall between the north trimmer and the south gable line. The answer turns on one thing the pair could not use: a single centered opening leaves two EQUAL segments, and the shear-wall engine derates the short one.

⛔ NOTHING ON THIS PAGE IS SELECTED. Register 1.2 records the R18 program as ruled; the EAST half is REOPENED (memo, same day) pending this study, and the SOUTH half stands confirmed. What survives the reopening unchanged is the exit role: register 1.2a migrated it on 2026-08-13, so the SOUTH SE-3 window is the second means of egress of record and the east opening is the deliberate below-reference supplementary escape. Its sill remains deliberately unset pending the PE’s fall-protection read (cover email §1.4). Every single window here is a dashed candidate with its arithmetic beside it.

⭐ REVISED 2026-08-13 — OWNER REVIEW, BEFORE PUBLICATION: THE THREE FINALISTS LEAD, EVERY DRAWN WINDOW CARRIES ITS SILL ABOVE THE FLOOR, AND THE HEAD-HEIGHT / RAFTER-TAIL VIEW SECTION IS NEW. EVERY OPTION PRICED BEFORE THE REVISION IS STILL PRICED, IN A DATED RECORD SECTION.

This study was written, reviewed by the owner and revised before it was ever published — the one legitimate reason to touch a frozen study, and the precedent is the 2026-08-10 wall-elevations study, revised the same way. The ruling it informs (R18-EAST) is still OPEN, so the page is still an input and not a record of a decision already taken.

⛔ What the revision did not do. It did not re-point the datum gate (still DATUM v9), did not re-derive a single figure, and did not delete an option: E-P, E-S2 and E-S4 keep their full rows, their full panels and their dots on the solve, in a dated record section. The 52.84 in maximum stays the labelled bound. ⭐ It tightened one thing: the workbook is now read from the committed blob of 20c39a6 rather than from the working tree — see the note at the foot of this page.

⭐ The rider, quoted rather than claimed. 2_LD_WIND §14.7 row 4, verbatim: “CHASE-SOUTH unchanged, but the torsion lands on the N-S lines: couple = 762 lb per N-S line. EAST's governing segment takes its length share and goes 0.451 -> 0.961 -- the E-W problem becomes an N-S problem.”

And R16, which changed what that rider costs: “Go with nailing upgrade” — the two N–S walls get the tighter Table 4.3A edge-spacing row, ADOPTED, with the spacing itself pending this study and the framing-footnote check. ⭐ The workbook has already bound it on WEST (sw__edge_sp_w reads 4 in) and not on EAST (sw__edge_sp_e still reads 6), in its own words: “the EAST instance’s bind travels with its window re-bind (its program is under revision) so that one commit carries the whole east state rather than leaving it half-migrated ”. That window re-bind is this ruling.

Basis: PE-package review docs/decisions/2026-08-13-pe-review-memo.md P1-1 · CALC: 2_LD_WIND §14.7 ‘THE E–W MECHANISM BRACKET’, row 4. Both cited because a utilization printed on a page owes its rationale and its live calculation, not one of them.

① The wall as R18 left it

At the same 40 inches of opening, one centered hole takes 13 % OFF the pair’s governing utilization — and the saving is exactly the ratio of the two aspect-ratio factors. The pair splits the wall 19.25 / 35.25; one centered window of the same width splits it 27.25 / 27.25. Σb is identical at 54.5 in, so the unit shear is identical too — but the pair’s short piece is derated to ARF 0.8214 while both balanced pieces sit at 0.9472, and the derated one governs. 1.159 → 1.005 under the couple; 0.795 → 0.689 with the R16 nailing. The generator asserts that the improvement equals the ARF ratio and nothing else — if anything else moved, the explanation would be wrong.

② The three finalists

finalist	segments	h/b	ARF	Σb	A — 50/50	B — couple	C — couple + R16 4 in
E-S1 — 40 × 30	27.25 + 27.25	2.42 / 2.42	0.947 / 0.947	54.5 in	0.472	1.005 ⛔	0.689
E-S1b — 40 × 24	27.25 + 27.25	2.42 / 2.42	0.947 / 0.947	54.5 in	0.472	1.005 ⛔	0.689
E-S3 — 32 × 24	31.25 + 31.25	2.11 / 2.11	0.986 / 0.986	62.5 in	0.396	0.842	0.577

⭐ E-S1 and E-S1b are one width row priced twice. Same 40 in of opening, so the same two 27.25 in segments, the same Σb, the same three utilizations and the same 44 in header span — the generator asserts that identity rather than trusting the re-computation to land on it. Everything that separates them is on figures ④ and ⑤.

Every cell re-runs the workbook’s own engine on its own transcribed constants (SDPWS Table 4.3A wind column 730 plf at the 6 in o.c. basis and 1065 plf at 4 in o.c., Table 4.3A fn.3 species factor 0.91 at G 0.41, the 4.3.3 ASD divisor 2, the 4.3.4.2 Aspect Ratio Factor, the Table 4.3.4 limit 3.5). Column A’s line shear is 675 lb ASD; column B adds the §14.7 couple of 762 lb to give 1,437 lb; column C changes capacity by the factor 1.4589 and changes Σb by nothing. Basis: owner rulings R7 / R16, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md · CALC: 9_TH_SW_E §I2 / §I3 and 4_ENG_SW’s Table 4.3A block.

③ How wide can one window be

⭐ The bound below is unchanged by the 2026-08-13 revision and is kept as the labelled bound it always was. What the revision did was name three finalists inside it; the curve, the 52.84 in crossing and the 56.79 in aspect cap are exactly as first derived.

The widest single window that closes is 52.84 in — and it is the mechanism that stops it, not the geometry. Two constraints could have bound. The aspect limit kills a segment below h/3.5 = 18.86 in, which would have allowed 56.79 in of opening. The §14.7 couple with the adopted R16 nailing crosses 1.000 first, at 52.84 in — 3.95 in short of it. Round down and 52 in is the widest whole inch that closes (0.971). ⚠ Without the R16 upgrade the same curve crosses at 39.77 in, so the adopted nailing is worth 13.06 in of window — and the pair’s own 40 in would have sat just outside it (0.23 in over). The solve is done in closed form and again by bisection on the drawn curve; the generator refuses the page unless the two agree to 1e−9.

④ ⑤ Windows onto rafter tails — the head-height question

⭐ The owner’s reading, 2026-08-13: “can we move two surviving options slightly below top plate; as currently stands it seems like these will be windows onto rafter tails…”

He is right, and it is the steep plane that does it. The east roof rises at 1.367 against the west’s 0.788, so outboard of this wall the eave drops away fast — and every head on figures ① and ② is drawn at the plate underside, which is the one place where the tail is directly overhead. The v8 taper lifts the tip back up: the tail underside at the 15 in eave tip derives to 49.37 in above the floor datum (⚠️ on EAVE_OVERHANG_NOMINAL — NOMINAL, FIELD VERIFY), against 43.44 without it.

The tails band is set by the HEAD alone. Width changes nothing; height changes only what fraction of the glass it is. At a given head and a given standing distance, every candidate on this page shows the same number of inches of tail — because the eave cuts the view at one height at the wall plane, and the head is the only thing above it. So the answer to “windows onto rafter tails” is the head, and only the head. At 2 ft back, dropping from 62 to 54 takes the tails band from 8.54 in to 0.54 in; at 1 ft it goes to nothing at all. And it is solved twice — a closed form and a bisection on a ray marched along the real tapered underside — which must agree before the page may be written.

head above floor	depth freed above the head	tails in view — 1 ft back	2 ft	3 ft	of E-S1’s 30 in glass (2 ft)	of E-S1b / E-S3’s 24 in (2 ft)
62 — plate underside	0 in	6.72 in	8.54 in	9.50 in	28 %	36 %
58	4 in	2.72 in	4.54 in	5.50 in	15 %	19 %
54	8 in	0.00 in	0.54 in	1.50 in	2 %	2 %

At a 60 in ILLUSTRATIVE eye — ⛔ no project source states an eye height; register 0.4’s convention is a 6 ft adult, a working stature and not a code figure, and the same-date porch-trunk-view study carries the same 60 in reference with the same disclosure. Standing distances are measured west of the wall line and stop at the ridge, because a standpoint further west looks out under the west plane, the ridge beam and the loft — none of which this section models. Geometry from geometry.py: TAN_E, EAVE_SEAT_E, EAVE_OVERHANG_NOMINAL, RAFTER_TAIL_TIP_D, TAIL_TIP_RISE_E, PLUMB_E — DATUM v9, taper rule DATUM v8 A/C.

A standing adult sees no open sky through this wall at all — and the hinge is the tail tip’s own height, from any distance. The ray that grazes the tip crosses the eye’s own level only when the eye is at 49.37 in, the height of the tip itself, and that is true at every standing distance — the distance cancels. So an eye above 49.37 sees only tail above its own level and only ground below it; an eye below it always keeps a band of open sky. A 60 in eye is above the hinge; a seated or a child’s eye is not. ⛔ What actually occupies the open air — sky, canopy, the oak, the meadow — is not modeled, and nothing here claims a view of anything. The level ray is drawn because it is exact: it separates looking up from looking down.

⚠ What a lower head costs, stated straight.

A lower head is a lower sill, inch for inch, at a fixed glass height. On this ladder the finalists’ sills run E-S1 32/28/24 · E-S1b 38/34/30 · E-S3 38/34/30 (inches above the floor, at heads 62/58/54). ⛔ Every one of them is below the 42 in kid band of register 0.3a — the highest sill any finalist reaches at any head on this ladder is 38, which is 4 in short of it. The sill stays deliberately unset pending the PE’s fall-protection read, exactly as the cover letter puts it; this page measures the gap and does not close it.
⭐ E-S1b is on the page for precisely this trade. It is E-S1 to the last digit in every structural column, and six inches shorter — so at every head it drops the head away from the soffit and keeps the sill six inches higher, at once. That is the one thing a shorter window buys, and it costs 240 in² of aperture.
A head below the plate underside needs short cripple studs between the header and the plate — ordinary framing, and nothing more. ⛔ This page sizes neither cripple nor header: no source states a header depth for these openings. What it can say is arithmetic: dropping the head frees 0/4/8 in above it, and the header tab’s own trial section is 7.25 in deep — so of the three, only head 54 leaves room for a section of that depth. ⭐ Read the other way round, the ladder is not a new ask at all: a real header under a 62 in plate underside puts the head near 54.75 anyway.
Shear does not see any of it. Every column in the table above is unchanged at every head — the segmented method counts full-height segments, so an opening costs its full width whatever its sill or its head. The generator re-prices every candidate at every head and refuses the page if a single figure moves.

Height, honestly: shear-free, but not free

⭐ Nothing in the shear arithmetic sees height. The segmented method counts full-height segments, so an opening costs its full width whatever its sill — every column above would be identical if each candidate were half as tall or twice. That is why the option set varies width and leaves height to the two lines that actually bound it.

⚠ But height is not header-free either, and it costs in two different directions.

WIDER costs a longer header. The header span is the opening plus the tab’s own bearing allowance (4 in) — ⚠️ not the opening width, which is what a first reading assumes. Across this whole option set the header’s governing check stays NOTCHED SHEAR @ combo 3 D+S and rises in a straight line with span: 32 in → 0.192; 40 in → 0.234; 48 in → 0.277; 52 in → 0.298. The tab’s own trial section (a doubled 2× at 3 × 7.25) would not reach 1.000 until a span of about 188 in — longer than the wall. ⭐ So no candidate here needs more than the doubled-2×8 basis, and the header is nowhere the binding constraint. ⛔ That is evidence about this class of opening on this wall, not a header specification for anything drawn here.
Basis: register 1.2a (R12/R13, exit role migrated by R18) · CALC: 9_TH_HDR_EXIT_E, an instance of 3_ENG_BEAM — re-run here at other spans and checked against its own four rows at its own span.
TALLER costs sill, and the header takes the top. Both height figures on figure ① are measured to the plate underside (62). A real bearing header comes out of the top of that: at the tab’s own trial depth of 7.25 in the head would land at 54.75, leaving 12.75 in above the 42 in kid band rather than 20. ⚠️ That is arithmetic on the tab’s trial section, not a stated depth — no source states one for these openings, and this page does not invent one. It is stated because a “20 in tall window” drawn to the plate underside is not a window anyone can build without answering it.

The two sill lines leave a two-inch band, and everything taller than 20 in is a request to the PE. Register 0.3a’s kid band puts an operable or breakable sill at 42; the owner-supplied reference basis caps an escape sill at 44 ([REG] REFERENCE, NOT ADOPTED — 0.1b). Between them the rough opening is 18 – 20 in tall, measured to the plate underside. E-S1’s 30 in implies a sill of 32 — 10 in below the band; E-S3’s 24 in implies 38. That is exactly the tension the letter already puts to the PE in its own words: “its sill is deliberately unset pending your read — our standing rule elevates operable glazing to 42 in for the kids, the egress function pulls the other way”. ⛔ This page measures it; it does not resolve it.

Aperture area — and what the escape role now asks of it

option	rough opening	implied sill	aperture area	vs the pair’s 896 in²	at the kid band (20 in tall)	header span · util
E-S1	40 × 30	32	1200 in² (8.33 sf)	1.34×	800 in² (5.56 sf)	44 in · 0.234
E-S1b	40 × 24	38	960 in² (6.67 sf)	1.07×	800 in² (5.56 sf)	44 in · 0.234
E-S3	32 × 24	38	768 in² (5.33 sf)	0.86×	640 in² (4.44 sf)	36 in · 0.192

Rough-opening areas, not glass: no frame or sash allowance is on file and no glazing product is selected, so no candidate can be said to meet or miss any figure. The reference basis on the record (owner-supplied with R12) is 20 in wide × 24 in high, 5.7 sf, 44 in maximum sill — carried as [REG] REFERENCE, NOT ADOPTED per 0.1b, and its minima do not combine into one opening. Note what the arithmetic says and no more: at the kid band, E-S2 gives 6.67 sf and E-S4 gives 7.22 sf, both above the 5.7 sf reference area while below its 24 in reference height — and both are now in the record section below.

⛔ Considered and set aside

CONSIDERED AND SET ASIDE (owner, 2026-08-13) — E-P, E-S2 and E-S4, with their full rows kept

The owner named E-S1, E-S1b and E-S3 as the finalists on 2026-08-13. Everything below was priced by this page before that and is priced by it still: same generator, same three columns, same datum, same figures. ⛔ It is moved out of the lead comparison, not deleted and not re-derived — a study is evidence of what was known and what was weighed, and dropping the options that were weighed would destroy exactly that.

set aside	rough opening	Σb	A — 50/50	B — couple	C — couple + R16 4 in	aperture	why it is here and not above
E-P	24×24 + 16×20	54.5 in	0.545	1.159 ⛔	0.795	896 in²	the reference the reopening replaced. It stays drawn in full on figure 1 and priced in full here, because the whole comparison is against it -- it is not a candidate and never was
E-S2	48 × 20	46.5 in	0.586	1.247 ⛔	0.855	960 in²	set aside on width: it needs column C to close at all, and column C's own footnote conditions are not on file
E-S4	52 × 20	42.5 in	0.666	1.417 ⛔	0.971	1040 in²	the bound itself, kept as the bound. 52.84 in remains the labelled maximum on figure 3 and the revision does not touch it -- what was set aside is BUILDING to it

⭐ What the role migration changes here. Register 1.2a migrated the second-means-of-egress role to the SOUTH SE-3 window on 2026-08-13; the east opening is the deliberate below-reference supplementary escape. So the east window is no longer the opening that has to answer a reference dimension — which is precisely why one larger hole is worth asking about: it can be sized on view, light and squeeze rather than on a minimum. R12’s own note on the record is that 24–30 in of width passes adults far more easily; every candidate here is at least 32 in wide. ⛔ That is the owner’s stated basis quoted — not a verdict on the adequacy of any candidate as an escape opening. The AHJ posture is one of the letter’s own open asks.

⛔ The ruling state, dated — and what is still one ruling behind

Ruled, then reopened, on the same day.

R18            2026-08-13  EAST = 24x24 centered + 16 in high band (contiguous 40 in)   RULED
R18-EAST       2026-08-13  "one larger e window rather than two on east"           REOPENED (same day)
1.2a           2026-08-13  exit ROLE migrates to SOUTH SE-3; east = supplementary  UNAFFECTED
R16            2026-08-13  EAST/WEST edge-nailing upgrade                          ADOPTED (spacing pends)

The register row 1.2 and the PE cover letter §1.4 both still carry the pair, because the reopening is recorded in the memo and has not been propagated. That is correct sequencing — the studies decide, then one commit propagates — but it means three documents will disagree until the east half is re-ruled, and this page exists to make that re-ruling decidable. ⛔ Nothing outside this folder is changed here.

And the workbook is still one ruling behind, now on two tabs. 2_LD_WIND’s dead2__east_opening_w is still 30 in — the superseded door reading — which feeds both 9_TH_SW_E’s Σb (64.5 in, two 32.25 in segments) and 9_TH_HDR_EXIT_E’s span (34 in). ⭐ This is held, not lost: the T11a build of the same day bound SOUTH and WEST and left the east state under a dated banner, in its own words because “the EAST instance’s bind travels with its window re-bind (its program is under revision) so that one commit carries the whole east state rather than leaving it half-migrated”. It is still why every guard on this page that verifies the engine runs against the workbook’s own 32.25 / 20 segments and its own 34 in span, while every guard that prices a candidate runs on the derived widths. ⛔ RAISED, NOT FIXED.
Basis: memo 2026-08-13-4x4-plate-and-nook-final.md §R18 and the same-day R18-EAST REOPENED entry · register 1.2 / 1.2a · CALC: 2_LD_WIND §13.1, 9_TH_SW_E §I2, 9_TH_HDR_EXIT_E inputs.

⚠ Where this arithmetic could be wrong, in either direction

In the wall’s favor — the engine is SEGMENTED-ONLY. 9_TH_SW_E’s SCOPE note: “SEGMENTED WALLS ONLY (SDPWS 4.3.5.1…). No perforated shear wall (4.3.5.3…) and no force-transfer-around-openings wall (4.3.5.2).” Both methods exist to credit sheathing above and below an opening — and a single large window leaves a continuous strip above and below it, which is exactly what they credit. ⭐ So these prices are the conservative ones, and a force-transfer treatment is the one route that could widen the window further. But not by an amount anyone may assume, and choosing a method is a PE-and-workbook decision, not a study’s.

Against it — this line’s share is held FIXED while its Σb changes. The SCOPE note also says “NO SHEAR-WALL-LINE DISTRIBUTION (4.3.3.4)”, so every candidate here changes this wall’s Σb and nothing else. A real distribution would move force between EAST and WEST as one of them is opened, and which way depends on the diaphragm.

And the couple column is not a ceiling. §14.7 brackets four mechanisms; the couple is the one that lands on this wall. Row 3 — two-support beam equilibrium, the row that satisfies moment equilibrium about a real point — puts CHASE-SOUTH at util 1.69 and drives SOUTH into reversal instead. The bracket is a band, not a worst case for every member at once.

And the R16 nailing is adopted, but its spacing is not. R16 takes the tighter Table 4.3A row as the design intent; which row (4 in or 3 in) is selected from this study’s own third column at the calcs bind, and the memo is explicit that the tight-spacing footnote is read from the on-file table “not assumed either way”. The workbook says the same about its own bind, and it is worth reading in full rather than in summary:

“MANDATORY CAVEAT, AND IT IS NOT A FORMALITY: Table 4.3A’s tighter-spacing rows carry FOOTNOTE CONDITIONS on framing width at abutting panel edges, fastener edge distance and panel layout, and THOSE FOOTNOTES ARE NOT TRANSCRIBED ANYWHERE ON FILE. This tab has bound the CAPACITY. It has NOT established that the wall can be BUILT to the row it now claims … If the footnote conditions are not met this capacity is not available and the wall reverts to the 6-in row.”

So column C’s ratio is the table’s and its buildability at this framing is not on file — and if it proves unavailable, every candidate here is read out of column B instead, where only E-S3 closes.
Quoted from 4_ENG_SW’s own R16 note, located by its row label on the recalculated workbook.

Deliberately NOT drawn

A choice, a recommendation, or an adopted width or height. The east half of R18 is reopened and this page is an input to re-ruling it.
A sill. The letter states it is unset pending the PE’s fall-protection read; this page draws the two lines that bound it.
A header depth or a header specification. The 7.25 in figure is the calc tab’s own trial section, used to show what a header would take out of the opening height — no source states one.
An adopted mechanism. The couple is the workbook’s row 4 of four, quoted; the ruling is the PE’s, over an open-front condition whose governing provisions are believed applicable and not transcribed.
A nailing SPACING, or any claim that the tighter row’s framing and edge-distance conditions are met.
Any glazing product, sash type, operation, opening-control device, hardware or glass specification, and no frame/sash allowance — none is on file, so no candidate can be said to meet or miss any reference figure.
Any code-compliance verdict. The SDPWS numbers are the workbook’s transcription; the utilizations are ours, not a licensed engineer’s. IRC-class figures appear only as [REG] REFERENCE, NOT ADOPTED (0.1b).
New wind arithmetic of any kind, and no utilization at any plate height other than R14’s ruled one.
A perforated or force-transfer-around-openings treatment. Named as the route that could widen the window, not run.
Hold-down hardware, and no deflection figure — 9_TH_SW_E’s is flagged NOT QUOTABLE.
Anything on the SOUTH wall. Its half of R18 stands confirmed and has its own study of the same date.
A HEAD HEIGHT. 62/58/54 is a ladder the owner asked to see, not a schedule. ⛔ Nothing on this page selects one.
An EYE HEIGHT as fact, a stature, or any child figure. The 60 in eye is ILLUSTRATIVE and labelled so everywhere it is drawn; the whole sweep is on the page.
Any claim about what is actually visible out of the window. The geometry says where the eave cuts the sightline; sky, canopy, the oak and the meadow are none of them in this datum.
A cripple-stud or header SIZE. A head below the plate underside needs short cripples — ordinary framing, named and not sized.
A re-derivation of anything the owner set aside. The record section carries the same figures the lead comparison carried before the revision.

References

(1) Code and standards — ANSI/AWC SDPWS-2015 ch.4: Table 4.3A (nominal unit shear, wind column, at 6 and 4 in o.c. panel edges; footnote 3 specific-gravity adjustment), 4.3.3 (ASD reduction factor 2), Table 4.3.4 (maximum aspect ratios; blocked wood structural panels 3.5:1), 4.3.4.2 (Aspect Ratio Factor = 1.25 − 0.125 h/b_s), 4.3.5.1 (segmented method), and the open-front / cantilevered-diaphragm provisions named — but not transcribed — by 2_LD_WIND §14.7. NDS-2018 as the header tab’s template applies it: 3.3.3 (beam stability, closed form), 3.4.2 / 3.4.3.2 (shear and notched shear), 3.10.2 / 3.10.3 (bearing), 3.5 (deflection, K_cr). ⛔ Every one is quoted here only as transcribed on 4_ENG_SW’s CODE BASIS block and in 3_ENG_BEAM’s template, which carry the printed page and the date it was read.
(2) Project sources of record — geometry.py DATUM v9; calcs.xlsx at 20c39a6 (the T11a state, post-aabca55), tabs 9_TH_SW_E, 9_TH_SW_W, 9_TH_HDR_EXIT_E, 2_LD_WIND §14.7, 2_LD_DEAD, 4_ENG_SW; docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R7 / §R12 / §R13 / §R14 / §R16 / §R18 and its same-day R18-EAST REOPENED entry; docs/decisions/2026-08-13-pe-review-memo.md P1-1; docs/decisions/2026-08-13-pe-cover-email.md §1.4 / §3; interiors register rows 0.1b / 0.1c / 0.3 / 0.3a / 1.2 / 1.2a / 1.5a.
(4) The eave, the tail taper and the sightline geometry (④, ⑤ — a NONSTANDARD detail drawn here, so it is cited specifically): the tapered eave tail is OWNER RULING 2026-08-12, DATUM v8, transcribed in geometry.py’s own banner — rule A (tapered on the UNDERSIDE beyond the plate’s outer face, to a RAFTER_TAIL_TIP_D 2 in tip measured square to the member, grading linearly to zero at the face), rule B (the plumb gain TAIL_TIP_RISE_E 5.928, derived in rederive()), rule C (nothing bearing-side moves) and ⚠️ rule D, still open: “[PE] eave snow/ice on the 2-in cantilevered tip — TO CONFIRM (tapered/plumb-cut tails are established framing practice, but that check is not on file)”. The soffit plane itself is geometry.py’s _soffit_e; the overhang is EAVE_OVERHANG_NOMINAL, sourced to 2.6a §6.4 J6 (a stated 12–18 in range) and NOMINAL — FIELD VERIFY. The same two functions on the WEST plane are drawn by the 2026-08-12-ladder-siting study, which is where the taper was ruled and which is frozen at v8. Register 0.4 supplies the 6 ft adult convention; ⛔ no source supplies an eye height, and this page invents none.
(3) Companion studies — 2026-08-13-east-window-program (frozen v9, the predecessor this page succeeds — its three published columns for the pair are asserted here), 2026-08-13-east-wall-openings (frozen v9), 2026-08-13-south-window-program (the south half of R18, which stands), 2026-08-10-wall-elevations (frozen v7).

Built from geometry.py at DATUM v9 (2026-08-13) — WALL_SPAN, TRIMMER_N_STATION, RIM_NS/S_CROSS_TO_S_RIM, PLATE_Z, PLATE_DEPTH, PLATE_W, PLATE_PLY, BOTTOM_PLATE, CHASE_E_WALL_X, CUTOUT_HEADER_W_X — and from calcs.xlsx at 20c39a6 (T11a, post-aabca55), read by defined name (and, for §14.7’s mechanism rows, 4_ENG_SW’s Table 4.3A spacing block and 9_TH_HDR_EXIT_E’s engine rows, which carry no defined names, by their own row labels) off a headless LibreOffice recalculation, never retyped. The generator asserts that its re-run reproduces 9_TH_SW_E’s published segments to 1e−6, 2_LD_WIND §14.7’s own couple row to the digit that row prints, 9_TH_HDR_EXIT_E’s sixteen engine cells to 1e−9 at its own span, and the predecessor study’s three published columns for the pair; that a centered single window leaves two equal segments priced identically; and that the maximum width solves the same from both directions. ⭐ Read from the COMMITTED BLOB of 20c39a6 (2cd1428a9ceb), extracted by git cat-file and SHA-pinned — not from the working tree. That changed during this revision, and for cause: this repo runs concurrent sessions, the calcs session was mid-flight on the east re-bind while this page was being revised, and the first REV run read a workbook whose east segments had already moved to 21.25 in and whose east edge spacing had already moved to 4 in. The guards caught it. A frozen study that reads a live file is not frozen. ⚠️ AT THIS BUILD THE WORKING TREE HAD MOVED AWAY FROM THAT COMMIT (it hashes f896e6ab75f9) — reported, deliberately not consumed, and not this page’s to reconcile. • FROZEN 2026-08-13, REVISED AND RE-FROZEN 2026-08-13 before publication. Do not regenerate. • This page is never to be measured from, and nothing on it is adopted.

Two windows on the south wall — above and below the loft

✅ RULED (R18, south half confirmed): the 27×32 egress slid hard east (best utilization, and the siting deleted a PE question) + a single 33×26 FIXED pane west of the ridge post (872 in², zero shear cost). The second-means-of-egress role lives here now.
⛔ FROZEN at datum v9 · 2026-08-13 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — NO WINDOW IS SELECTED, NO SIZE OR STATION IS ADOPTED, NO CODE FIGURE IS ADOPTED AND NO MECHANISM IS ADOPTED — GENERATED FROM geometry.py AND A RECALCULATED calcs.xlsx, NOT HAND-DRAWN

Two windows on the south wall — above and below the loft

One wall, cut in two by the loft, with opposite economics either side of the top plate. Above it the gable triangle is not part of the shear wall at all, so a window there costs zero Σb. Below it is a counted E–W shear line at 0.293, and that is the margin the ruled egress window is spent out of. Above the loft the question is shape; below it the question is width, and where along the wall.

⛔ NOTHING ON THIS PAGE IS SELECTED. Register 1.2 is OPEN. 1.2f rules the requirement — one window in the south wall opposite the north door, sized for emergency exit — and rules nothing about its size or station. 1.2c (the gable pair) is OPEN, and the owner’s program of one window above the loft is unruled in count as well as in size. Every opening on this page is a dashed candidate with its arithmetic beside it.

① The whole wall, both windows

Above the loft, the biggest single pane is not the most glass. The ridge post splits the gable into unequal halves: a 66.5 in west run and a 37.5 in east one. The best single rectangle on the west side is 33.25 × 26.21 — 872 in², 1.81× the best one on the east. But register 1.2c’s pair, each at its own peak, is 961 in² of glass — 1.10× the single. One big window or two smaller ones is a look decision, and both are free.

② Above the loft

⚠ An above-loft window is at KID HEIGHT for anyone ON the loft — and nobody has ruled what that means. Every gable rectangle sits on the plate line, 66 in above the room’s floor. The loft deck is at 54, so the same sill is 12 in above the floor a child is actually on. Register 0.3a governs glazing that can open or break and sets the band at 42 in — but it does not say from which floor.

Read from the room floor, the whole gable is above the band and every option is available.
Read from the loft floor, the band lands 30 in above the plate, and the free window shrinks to 14.22 × 11.21 west (159 in², 18% of G-A) or 7.78 × 10.63 east.

⛔ This page does not rule it and could not. It draws both readings, on the same triangle, so the question can be asked with a number attached. It is also not the only thing unstated up there: what a loft occupant can reach and whether fixed glazing changes the answer are 0.3a’s to say, and the fall-protection posture for operable openings is already an open PE item on the east wall (cover email §1.4).

③ Below the loft — the ruled window

Sliding the window EAST along the ruled stretch is free, and it is never worse. Hard against the ridge post the window leaves a 14.5 in stub that stops counting — Σb 66.5 in, util 0.476. Slid to the east end it leaves one uninterrupted 81 in segment at util 0.391. The generator sweeps every width from 0.57 in to the full stretch and asserts the east-most position never governs worse than the west-most one — it is a property of the geometry, not of one candidate. ⚠️ And it is not free of questions: the east jamb lands at the building corner, and no source details the corner framing or the hold-down anchorage there. Nothing is invented for it here.

⭐ More Σb can be worse, and this wall shows it. SE-4 is narrow enough (22.5 in, inside the 22.64 in ceiling) to keep the east stub alive at 19 in. That buys it 85.5 in of Σb against SE-3’s 81 — and it still governs worse (0.454 against 0.391). A segment at h/b 3.474 is derated by the 4.3.4.2 Aspect Ratio Factor to 0.816, so it becomes the weakest link even though it adds length. Σb is a sum; utilization is governed by a member. The generator asserts the inversion rather than asserting at it.

④ The ledger

candidate	full-height segments	Σb counted	v (plf)	governing util	what it says
as it stands — no opening	108	108 in	97.4	0.293	the R10 two-line state this window is spent out of
SE-1 — 27 × 32, hard WEST (X 66.5)	66.5 + 14.5	66.5 in	158.2	0.476	the companion study's position -- and its west jamb lands ON the loaded S gable ridge post. The east stub dies
SE-2 — 27 × 32, CENTERED in the stretch	73.75 + 7.25	73.75 in	142.7	0.430	clear of the post, and the stub it leaves is smaller still
SE-3 — 27 × 32, hard EAST (jamb at X 108)	81	81 in	129.9	0.391	one uninterrupted segment and no stub to lose -- but the jamb lands at the building corner, which no source details
SE-4 — 22.5 × 32, hard WEST — inside the width ceiling	66.5 + 19	85.5 in	123.1	0.454	narrow enough to keep the east stub alive at 19.0 in -- and that stub is derated so hard it governs WORSE than SE-3's single long segment
any ABOVE-LOFT window	108	108 in	97.4	0.293	⭐ unchanged — the gable is not part of the shear wall

Every row re-runs 9_TH_SW_S’s own engine on its own transcribed constants (SDPWS Table 4.3A wind column 730 plf at the adopted 6 in o.c. panel edges, Table 4.3A fn.3 species factor 0.91 at G 0.41, the 4.3.3 ASD divisor 2, the 4.3.4.2 Aspect Ratio Factor, the Table 4.3.4 limit 3.5). The generator refuses to write this page unless the un-opened wall reproduces the workbook’s own v, v_allow, h/b and utilization to 1e−6. Basis: owner ruling R10, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R10 · CALC: 9_TH_SW_S, LIVE headline / §I1. Both cited because a utilization printed on a page owes its rationale and its live calculation, not one of them.

⚠ What the mechanism bracket does to THIS wall

The couple leaves SOUTH alone. The bracket does not. On 2026-08-13 the PE-package review published an E–W mechanism bracket (2_LD_WIND §14.7): the adopted 50/50 split is a placeholder and the mechanism is a PE ruling, over an open-front condition whose governing provisions are believed applicable and are not transcribed. Four candidates, and what each leaves for this wall:

mechanism (§14.7 row)	V to CHASE-SOUTH, ASD	V to SOUTH, ASD	SOUTH util	what it means here
1. ADOPTED WORKING 50/50	877 lb	877 lb	0.293	leaves 6,856 lb-ft of torsion UNPLACED -- no element is assigned to resist it
2. POSITION-BASED TRIBUTARY	1,452 lb	302 lb	0.101	SOUTH keeps the remainder; statics close in shear but the torsion is still carried by the eccentricity term, not by an assigned element.
3. TWO-SUPPORT BEAM EQUILIBRIUM	2,547 lb	-793 lb ⛔ REVERSAL	0.265	SOUTH GOES INTO REVERSAL: R_south = -793 lb, i.e. it must be ANCHORED AGAINST UPLIFT-SIDE REVERSAL rather than merely resisting shear. No device on this project is checke
4. ADOPTED 50/50 + the residual torsion resolved as	877 lb	877 lb	0.293	CHASE-SOUTH unchanged, but the torsion lands on the N-S lines: couple = 762 lb per N-S line. EAST's governing segment takes its length share and goes 0.451 -> 0.961 -- th

⭐ Row 4, the COUPLE, is the one that matters on the EAST wall and does nothing here. It leaves CHASE-SOUTH exactly where the adopted split has it and resolves the residual E–W torsion as a couple of 762 lb per line into the north–south walls. SOUTH is an east–west line: it is not in that load path, and its utilization does not move by a digit. (The companion east-wall study of this date prices exactly what it does over there.)

⛔ But SOUTH is not unexposed to the bracket — and its exposure is a change of SIGN, not of size. Every other mechanism takes SOUTH’s shear down. Row 3, the only one that satisfies moment equilibrium about a real point with the load where it actually is, takes it past zero: -793 lb. In the workbook’s own words the wall “must be ANCHORED AGAINST UPLIFT-SIDE REVERSAL rather than merely resisting shear. No device on this project is checked for that.” A bigger window does not cause that and a smaller one does not prevent it — but it is the reason the comfortable 0.293 on this page is not the whole story about this wall.
Basis: PE-package review

docs/decisions/
2026-08-13-pe-review-memo.md

P1-1 · CALC: 2_LD_WIND §14.7 ‘THE E–W MECHANISM BRACKET’, rows 1–4, read by row label. Band in one line: “E-W split to CHASE-SOUTH, ASD: length-proportional 588 lb (floor) | ADOPTED 50/50 877 lb -> util 0.58 | position tributary 1,452 lb -> util 0.96 | beam equilibrium 2,547 lb -> util 1.69 (ceiling). THE ADOPTED VALUE IS THE SECOND LOWEST OF FIVE AND THE MECHANISM IS UNRULED.”

What the window runs into — raised, not designed

⭐⭐ The ruled stretch’s west edge IS a loaded post. The S gable ridge post stands in this wall at X 66.5–70.5 — exactly where the door-opposite stretch begins — running floor to the ridge-beam soffit and carrying 271.6 lb. SE-1’s west jamb lands on it. ⛔ Whether a loaded ridge post may also serve as a window king stud is a PE question and it is not answered here — it is already in the letter. ⭐ Note what the east-most siting does to it: it removes the question, because the window no longer touches the post. That is a second, independent argument for sliding east, and it is not this page’s to weigh against the corner-framing one.
Basis: 2.6a §0 / LIVING §12 · CALC: 9_TH_RIDGE_3SPAN §5 ‘SUPPORT REACTIONS’. The seat below the floor is already scheduled J12, PE TO SPECIFY; the king-stud question is in the 2026-08-13 cover email §1.4.
⭐ A ROUGH opening is not a CLEAR one, and the difference is not on file. The letter’s figure is 27 × 32 CLEAR. Frame and sash come out of a rough opening and no source on file states that allowance, so the candidates here are drawn at the clear figures and every Σb price is a lower bound. ⭐ What survives that uncertainty is the ranking: the sweep above holds at every width, so a larger rough opening moves the numbers and not the order.
The loft ledger hits every stud on this wall, and an opening interrupts it. Over the opening’s width the intermediate studs are cut, so only the jamb studs are behind the ledger there. ⛔ No source states how the ledger is carried across an opening — flagged for the ledger detail, not invented (register 2.1, still OPEN).
The kid-safe sill band is deliberately broken below the loft, and that is 0.3’s own exception. Every candidate’s sill is at 19 in, below the 42 in line and below the [REG] reference maximum of 44. An emergency exit you cannot climb to is not one.
Headers are cheap in this building, and that is measured, not assumed. The north door’s lintel runs at 0.014 — NOTCHED SHEAR @ combo 1 D. ⛔ That is evidence, not a header size. No source states a header depth for these openings, so every head here is drawn at the loft frame’s underside with the header still to come out of it.
CALC: 9_TH_HDR_DOOR, GOVERNING UTILIZATION.

⭐ The predecessors’ raise LANDED — recorded, because a protocol that works should be visible too

The clear framed height between plates is 60.5 in at DATUM v9, not the 61.5 in the project published until this date (PLATE_Z 66 − PLATE_DEPTH 4 − BOTTOM_PLATE 1.5). Datum v9 replaced the doubled 2× top plate with a single rough true 4×4 and its banner says where the inch went: “the extra inch of plate depth comes out of the studs.”

The two 2026-08-13 opening studies raised it, changed nothing outside their own folders, and left the owning documents to the owner. As of this page all three live register rows carry the v9 figure with a dated supersession: 1.5a 61.5 → 60.5; 0.3 21 → 20 in of rough opening above a 42 in sill; 1.3b 10.5 → 11.5 in of overrun. ⭐ That is the discrepancy protocol closing a loop, and it is worth a line: a study reports, the owner rules, and the owning document is fixed in its own commit.

⭐ It never moved this page’s below-loft arithmetic in any case: that zone is bounded by the loft frame (51 in), not by the top plate. ⛔ Nothing outside this folder was changed here either.

Deliberately NOT drawn

A size, a station or a count for either window.
Any adoption of a code figure. IRC-class figures appear only as [REG] REFERENCE, NOT ADOPTED: 0.1b rules egress sizing a research gate, 0.1 records that the figures were quoted from memory, and whether the provision applies at all is an AHJ question.
Any code-compliance verdict, for the window or the wall.
Any frame, sash, hardware, operation, opening direction, glazing product or glass specification — and no rough-to-clear allowance, because none is on file.
Any header depth, above or below the loft.
Any king-stud, jamb or post detail at X 66.5 — that is the PE question this page carries.
Any corner framing or hold-down detail at the east jamb of the east-most siting. It is named as the cost of that siting, not designed.
A ruling on whether 0.3a’s band is measured from the loft floor, or on what a loft occupant can reach.
The loft ledger’s detail across an opening, or any blocking layout (register 2.1 is open).
Any vent area, vent sizing rule or venting scheme — 3.3a is a shut research gate.
An adopted mechanism, and any new wind arithmetic — including redistribution between the two E–W lines and any utilization at a 72 in plate, which R14 has in any case closed.
A perforated or force-transfer-around-openings treatment. The engine is segmented-only and says so, which makes these prices the conservative ones — but not by an amount anyone may assume.
Hold-down hardware. The tensions are demands; no device is selected anywhere in this project, which is why 9_TH_SW_S’s deflection is flagged NOT QUOTABLE and is not quoted.
Anything on the EAST wall — the two-window east program has its own study of the same date.

References

(1) Code and standards — ANSI/AWC SDPWS-2015 ch.4: Table 4.3A (nominal unit shear, wind column; footnote 3 specific-gravity adjustment), 4.3.3 (ASD reduction factor 2), Table 4.3.4 (maximum aspect ratios; blocked wood structural panels 3.5:1), 4.3.4.2 (Aspect Ratio Factor = 1.25 − 0.125 h/b_s), 4.3.5.1 (segmented method), and the open-front / cantilevered-diaphragm provisions named — but not transcribed — by 2_LD_WIND §14.7. IRC R310 (‘Emergency Escape and Rescue Openings’) appears as [REG] REFERENCE, NOT ADOPTED only, with its edition-dependence and applicability both unresolved. ⛔ Every code number is quoted only as transcribed on the workbook’s own CODE BASIS block, which carries the printed page and the date it was read.
(2) Project sources of record — geometry.py DATUM v9; calcs.xlsx post-aabca55, tabs 9_TH_SW_S, 9_TH_SW_CS, 9_TH_HDR_DOOR, 2_LD_WIND §14.7; docs/decisions/2026-08-13-pe-cover-email.md §1.4; docs/decisions/2026-08-13-pe-review-memo.md P1-1; docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R10 / §R14; interiors register rows 0.1b / 0.3 / 0.3a / 1.2 / 1.2c / 1.2d / 1.2f / 1.5a / G-1 / 2.1.
(3) Companion studies — 2026-08-13-south-wall-openings (frozen v9, the predecessor this page succeeds), 2026-08-13-east-window-program (the east pair, same date), 2026-08-13-east-wall-openings, 2026-08-10-wall-elevations (frozen v7).

Built from geometry.py at DATUM v9 (2026-08-13) — WALL_SPAN, RIM_NS/S_CROSS_TO_S_RIM, PLATE_Z, PLATE_DEPTH, BOTTOM_PLATE, RIDGE_X, POST_W, RIDGE_ARRIS_W/_E, APEX_RISE, TAN_W/TAN_E, LOFT_TOP_Z, LOFT_FRAME_D, BELOW_LOFT_CLEAR, ABOVE_NET_AT_EAVE, LOFT_STRUCT_EW, WALL_STUD_THK, Z_RIDGE_BOT/_TOP, RIDGE_POST_FLOOR_TO_SOFFIT, RIDGE_POST_LOAD_S_GABLE_LB — and from calcs.xlsx post-aabca55, read by defined name (and, for §14.7’s mechanism rows, which carry none, by their own row label) off a headless LibreOffice recalculation, never retyped. The generator asserts that G-1’s triangle closes on the plate at exactly 0.0000 at both wall lines, that every gable rectangle sits wholly above the plate and under the slope, that the un-opened wall reproduces 9_TH_SW_S’s own published figures to 1e−6, that the east-most siting never governs worse than the west-most one at ANY width, and that no drawn label prints the superseded 61.5 in clear height or the rafter-plane apex G-1 forbids. • FROZEN 2026-08-13. Do not regenerate. • This page is never to be measured from, and nothing on it is adopted.

Two windows on the east wall — the pair (superseded record)

⏹ Ruled, then REOPENED the same day — the owner chose one larger window instead; the single-window study above is the successor. Kept as the record: the pair's mechanism arithmetic and the finding that only the narrow second window closed under the couple.
⛔ FROZEN at datum v9 · 2026-08-13 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — NO WINDOW IS SELECTED, NO SIZE IS ADOPTED, NO SILL IS SET, NO MECHANISM IS ADOPTED AND NO NAILING SCHEDULE IS ADOPTED — GENERATED FROM geometry.py AND A RECALCULATED calcs.xlsx, NOT HAND-DRAWN

Two windows on the east wall — and the mechanism rider

The owner wants two windows on the east wall. One of them is ruled: R12 makes the second exit an emergency squeeze window and R13 sites it centered. The other is not ruled at all. This page prices the pair — and it prices them three times, because on 2026-08-13 the PE-package review published a mechanism bracket in which this wall’s governing segment goes from 0.451 to 0.961 without a single opening being cut.

⛔ NOTHING ON THIS PAGE IS SELECTED. Register 1.2 — the window schedule — is OPEN. What is ruled on this wall is 1.2a-rev: the second exit is an emergency squeeze window (R12, working figure 24 × 24) sited CENTERED (R13). Its size is a working figure and its sill is deliberately unset pending the PE’s fall-protection read (cover email 2026-08-13 §1.4). ⛔ The SECOND window is UNRULED ENTIRELY — it appears in no register row and no memo, and everything this page says about it is a dashed candidate with its arithmetic beside it.

⭐ The rider, quoted rather than claimed. 2_LD_WIND §14.7 row 4, verbatim: “CHASE-SOUTH unchanged, but the torsion lands on the N-S lines: couple = 762 lb per N-S line. EAST's governing segment takes its length share and goes 0.451 -> 0.961 -- the E-W problem becomes an N-S problem.”

And the band it sits in, wind__ew_bracket: “E-W split to CHASE-SOUTH, ASD: length-proportional 588 lb (floor) | ADOPTED 50/50 877 lb -> util 0.58 | position tributary 1,452 lb -> util 0.96 | beam equilibrium 2,547 lb -> util 1.69 (ceiling). THE ADOPTED VALUE IS THE SECOND LOWEST OF FIVE AND THE MECHANISM IS UNRULED.”

Basis: PE-package review docs/decisions/2026-08-13-pe-review-memo.md P1-1 · CALC: 2_LD_WIND §14.7 ‘THE E–W MECHANISM BRACKET’, row 4. Both cited because a utilization printed on a page owes its rationale and its live calculation, not one of them.

① The wall as ruled — and a sill that will not fit

At DATUM v9 a 2′×2′ window cannot sit above the kid-safe sill band. There are 20 inches up there, not 24. The plate underside is 62 and register 0.3a’s band starts at 42, so the rough opening available above it is 62 − 42 = 20 in. A 24 in tall opening on that sill would reach 66 — 4 in inside the top plate. So the owner’s square and the kid band are mutually exclusive by 4 inches, and that is exactly the tension the PE letter already puts on the table: “its sill is deliberately unset pending your read — our standing rule elevates operable glazing to 42 in for the kids, the egress function pulls the other way.” ⛔ This page does not resolve it. It measures it.

⚠ And the 20 is itself a corrected figure — corrected because the predecessor study raised it. Register 0.3’s reconciliation used to publish “21 in of rough opening” above a 42 in sill, on the v8 doubled plate. At DATUM v9 the single rough true 4×4 plate took an inch out of the studs, so it is 20.

⭐ That correction has LANDED, and it is worth recording that the protocol worked. The two 2026-08-13 opening studies raised the clear framed height as 60.5, not the v8 61.5, and changed nothing outside their own folders. As of this page all three live rows carry the v9 figure with a dated supersession: 1.5a 61.5 → 60.5, 0.3 21 → 20, 1.3b 10.5 → 11.5. ⛔ Nothing outside this folder was changed here either — the raise on THIS page is the workbook one below, and it is still open.

⛔ The workbook is one ruling behind — RAISED, NOT FIXED

9_TH_SW_E still binds the superseded ~30 in DOOR, not R12’s ~24 in squeeze window.

wind__len_east_net  =  94.5 − 30  =  64.5 in   →  two 32.25 in segments   (the DOOR reading, still live)
R12 / R13           =  94.5 − 24  =  70.5 in   →  two 35.25 in segments   (the RULED window, centered)

R12 was ruled on 2026-08-13 and the T10 calcs run of the same day did not carry it into dead2__east_opening_w. The consequence is in the wall’s favor — a smaller opening leaves 6 in more Σb — so nothing published is unsafe. But two documents now state different segment lengths for one wall, which is what the discrepancy protocol exists for. ⛔ This page derives the R12/R13 state and RAISES the disagreement; it does not fix the workbook. Every guard on this page that verifies the engine is run against the workbook’s own 32.25 / 20 segments and reproduces its published utilizations to 1e−6; every guard that prices a candidate is run on the ruled 35.25 in.
Basis: memo 2026-08-13-4x4-plate-and-nook-final.md §R12 / §R13 · register 1.2a-rev · CALC: 2_LD_WIND §13.1 and 9_TH_SW_E §I2.

② Four combinations

③ The argument

combination	full-height segments	Σb	A — 50/50 baseline	B — §14.7 couple	C — couple + 4 in o.c.
E-0 — ONE WINDOW — the ruled squeeze window alone	35.25 + 35.25	70.5 in	0.346	0.736	0.505
E-1 — + SECOND WINDOW 24 in, HIGH BAND, NORTH	11.25 + 35.25	35.25 in	0.692	1.473 ⛔	1.009 ⛔
E-2 — + SECOND WINDOW 24 in, HIGH BAND, SOUTH	35.25 + 11.25	35.25 in	0.692	1.473 ⛔	1.009 ⛔
E-3 — + SECOND WINDOW 16 in, HIGH BAND, NORTH	19.25 + 35.25	54.5 in	0.545	1.159 ⛔	0.795

Every cell re-runs the workbook’s own engine on its own transcribed constants (SDPWS Table 4.3A wind column 730 plf at the adopted 6 in o.c. and 1065 plf at 4 in o.c., Table 4.3A fn.3 species factor 0.91 at G 0.41, the 4.3.3 ASD divisor 2, the 4.3.4.2 Aspect Ratio Factor, the Table 4.3.4 limit 3.5). Column A’s line shear is 675 lb ASD; column B adds the §14.7 couple of 762 lb to give 1,437 lb; column C changes capacity by the factor 1.4589 and changes Σb by nothing. The generator refuses to write this page unless its re-run reproduces 9_TH_SW_E’s published segments and 2_LD_WIND §14.7’s own couple row. Basis: owner ruling R7, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R7 · CALC: 9_TH_SW_E §I2 / §I3.

The two-window program is hostage to a ruling nobody has made yet — unless the nailing rides with it. The one-window state survives the couple comfortably (0.736). Every two-window combination fails it. The 4 in edge nailing rescues exactly one of them — the narrow second window, 1.159 → 0.795 — and at 24 in it is not enough (1.009, over by 0.009). A capacity upgrade multiplies what is left; it cannot buy back a segment that has stopped counting.

What the three columns are, and what they are not

A — the 50/50 baseline. The N–S split the workbook publishes: two counted lines at the two edges of the diaphragm, half the base shear each (owner ruling R7). 675 lb ASD to this line.
Basis: owner ruling R7, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R7 · CALC: 9_TH_SW_E §I2 / §I3.
B — the §14.7 couple. Not a new analysis: the fourth of four mechanisms the workbook itself brackets for the E–W direction. Under it CHASE-SOUTH is unchanged and the residual E–W torsion is resolved as a couple into the N–S walls — 762 lb per line, which on this wall is larger than its entire direct share. ⚠ The E–W problem becomes an N–S problem, in the workbook’s own words. ⛔ No mechanism is adopted here. §14.7’s own banner says it: “THE MECHANISM IS A PE RULING, AND NOTHING HERE ADOPTS ONE”, over an open-front condition whose governing provisions are believed applicable and not transcribed.
Basis: PE-package review docs/decisions/2026-08-13-pe-review-memo.md P1-1 · CALC: 2_LD_WIND §14.7 ‘THE E–W MECHANISM BRACKET’, row 4. Both cited because a utilization printed on a page owes its rationale and its live calculation, not one of them.
C — tighter panel-edge nailing. The same Table 4.3A row at 4 in o.c. panel edges instead of the adopted 6 in: nominal wind unit shear 1065 plf against 730, a factor of 1.4589. It is on file precisely for this — 4_ENG_SW’s own note says the tighter rows were transcribed “so an instance that needs more capacity can move up the row without re-opening the standard”. It costs no Σb and no window.
⛔ What this page may NOT say about it. Table 4.3A’s tighter rows carry conditions — framing width at abutting panel edges, nail edge distance, panel layout — and those footnotes are not transcribed on 4_ENG_SW. Only the capacity numbers are. So the ratio is the table’s and the permissibility is not on file: tighter nailing is priced here as a capacity lever and is not adopted, not scheduled and not certified buildable at this framing. That is a research-and-PE item, and inventing the conditions would be exactly the failure the never-invent rule exists to stop.

⭐ What the combinations actually say

The budget on this wall is 16.39 in, and 18 in is already over it. Each segment the ruled window leaves is 35.25 in. A segment stops counting below h/3.5 = 18.86 in, so a second opening can take 35.25 − 18.86 = 16.39 in and no more. At 24 in the host segment leaves 11.25 in and stops counting entirely — one opening, one whole segment.
North or south is free, and that is worth knowing. E-1 and E-2 are mirror images and they are identical in every column — same Σb, same unit shear, same governing utilization, to the last digit the generator can hold. Nothing on this wall is asymmetric. So the side is chosen on view, light, furniture and the ladder arrival, not on shear, and the generator asserts the equality rather than asserting at it.
The narrow second window is the only one that survives — and only with the nailing. E-3’s 16 in window leaves 19.25 in, which counts at h/b 3.429 but is derated hard (ARF 0.821). At the baseline it is comfortable (0.545); under the couple it fails (1.159); with the tighter nailing it passes (0.795).
Headers are cheap on this wall, and that is measured, not assumed. The ruled exit’s header runs at governing utilization 0.181 — NOTCHED SHEAR @ combo 3 D+S. ⛔ That is evidence about this class of opening on this wall, not a header size for anything drawn here: no source states a header depth for these openings, so every rough opening runs to the plate’s underside with the header still to come out of it.
Basis: register 1.2a-rev (R12/R13, 2026-08-13) · CALC: 9_TH_HDR_EXIT_E, GOVERNING UTILIZATION / §HDE1.

⚠ Where this arithmetic could be wrong, in either direction

Deliberately NOT drawn

A choice, a recommendation, or an adopted size for either window. Register 1.2 is open and this page is an input to it.
A sill for the ruled window. The letter states it is unset pending the PE’s fall-protection read; this page draws the two lines that bound it and measures the gap between them.
An adopted mechanism. The couple is the workbook’s row 4 of four, quoted; the ruling is the PE’s.
An adopted nailing schedule, and any claim that the tighter Table 4.3A row’s framing and edge-distance conditions are met — they are not transcribed anywhere on file.
Any header depth, for either opening.
Any glazing product, sash type, operation, opening-control device, hardware or glass specification, and no frame/sash allowance — none is on file, so no candidate can be said to meet or miss any reference figure.
Any code-compliance verdict. The SDPWS numbers are the workbook’s transcription; the utilizations are ours, not a licensed engineer’s. IRC-class figures appear only as [REG] REFERENCE, NOT ADOPTED (0.1b).
New wind arithmetic of any kind — including a utilization at a 72 in plate, which R14 has in any case closed.
A perforated or force-transfer-around-openings treatment. Named, not run.
Hold-down hardware. The tensions are demands; no device is selected anywhere in this project, which is why 9_TH_SW_E’s deflection is flagged NOT QUOTABLE and is not quoted.
Venting. Register 3.3a is a shut research gate, and a high band that opens is a vent.
Anything on the SOUTH wall — the two-window south program has its own study of the same date.

References

West-mid support — three ways to attach to the oak

⏸ Open. All three live; PE + arborist gated, supplier consult pending.
⛔ FROZEN at datum v7 · 2026-08-09 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — ALL THREE ARE LIVE, NONE IS CHOSEN — PE + ARBORIST GATED — GENERATED FROM geometry.py

Three ways to hang the west beam off the oak

The west beam needs one intermediate support. The first choice was a sixth screw pile, and it is root-blocked — a large primary anchoring root we are not willing to cut. So the remaining three all touch the tree, and the question is what each one does to it.

What the tree is being asked to carry. The support takes 1,982 lb with the beam spliced (2,476 lb if it runs continuous), at a reach of 24.5 in from the trunk face. That reach turns the load into a 4,047 ft-lb twist, and a bracket cannot hold a twist at one point — it needs two attachments, one pulling out and one pushing in.

⭐ The single most useful number here

Spreading the two anchors further apart is free, and it is the whole game.

Anchors spaced	Load in EACH anchor
24″ apart	2,000 lb
36″ apart	1,350 lb	⭐ a third less, for nothing but placing the bolts further apart

Both spacings are recovered from the published couple (M × 12 / P), not typed — if the published set ever stops closing, this study refuses to print a spacing nobody stated. Ask the supplier how much further the spacing can keep helping.

① CONFIG 1 — three-member right triangle, west face

OUR OWN PROPOSAL — NO SUPPLIER PRODUCT

Right angle at the TOP. Both tree bolts on one vertical member.

A vertical piece lies against the trunk with both tree connections on it, one at each end. An arm runs from its top out to the beam, and a diagonal closes the triangle back down to its foot. The diagonal pushes the lower bolt in; the arm pulls the upper bolt out — and pulling out is how a lag screw leaves a tree, which is why the upper anchor governs.

⛔ Do not confuse this with ② below. This one is apex-UP with the right angle at the top; the Tribeam is apex-DOWN. They have been mixed up before, in both directions. ⚠️ And this is the one option no product photo governs — it is our sketch, so a supplier must price and bless it rather than recognize it.

② CONFIG 2 — the Tribeam, south face, spanning both beams

SUPPLIER PRODUCT — DRAWN FROM THE PHOTOS ON FILE

Apex-DOWN: one long member across the top, two struts converging to a low anchor.

The biggest of the three, and the only one that touches both beams. A high tab carries a member across the top; two diagonals run down and inward to a single low anchor.

⛔ The trunk is not centered between the beams — 37″ to the west beam against 71.75″ to the east, nearly 2:1. A symmetric product on an asymmetric span is the open question, and it is the first thing to put to the supplier. ⚠️ No source publishes how this one splits its anchor loads, so this study says only what the geometry forces and leaves the spacing undimensioned rather than inventing one.

③ CONFIG 3 — knee-brace bracket

SUPPLIER PRODUCT — DRAWN FROM THE PHOTOS ON FILE

Arm at beam level; diagonal runs DOWN to a lower anchor.

The simplest. An arm bolts to the trunk at beam height and picks the beam up; a diagonal runs from the beam underside down to a second, lower trunk bolt. Because that diagonal is in compression — a prop, not a hanger — this option props the beam rather than hanging it, and the tree carries less than it does under ①.

Where this leaves it

	① right triangle	② Tribeam	③ knee brace
Face of the tree	west	south	west
Beams served	west only	both	west only
Is it a product?	⚠️ no — ours	yes	yes
Anchor loads published	yes — out/in split	⚠️ no	⚠️ not for this arrangement
Load on the tree	pull-out governs	consultant to confirm	⭐ props — least

What no drawing can settle. All three are PE + arborist gated, and every member except this one support touches no tree at all. Bracket hardware above is schematic and not to scale — brackets sized for nominal stock will not fit true-4″ rough hemlock, which is its own question for the supplier. The technical comparison, at one scale with all eight views, is S-104 in the construction set; this study is the plain-language companion to it, not a replacement.

The west nook — bookshelf, or nothing?

⏸ What it holds is still open (3.7 / 3.10) — the geometry has since been ruled past this study. Its 6-in-usable figure was SUPERSEDED the same day (2.2a-rev: the 12-in sway allowance is an EAST-side constant; the nook IS the 12-in west short-joist bay, width ruled 1 ft, and it has floor), and depth was RULED at the 26.25-in maximum (2.2b, 2026-08-13). Kept as the record of the question.
⛔ FROZEN at datum v7 · 2026-08-10 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — GENERATED FROM geometry.py, NOT HAND-DRAWN — DATUM v7, FROZEN 2026-08-10

The west nook — it is the west short-joist bay

The space between the west cabin wall and the oak has been carried as “the book nook” since the loft memo adopted it, with no dimension attached to it anywhere. It has one, and it is not a leftover: the nook is a floor framing bay, bounded by the west beam on one side and the west double header on the other.

1'-0" wide, west beam CL to the west double header — and that is a bay, not a clearance.

The datum reaches the same 12.0 by two routes that are the same route: CUTOUT_HEADER_W_X = TRUNK_W_FACE_X 18 − TRUNK_STANDOFF 6, and SHORT_JOIST_SPAN_W, the span of the short joists that carry the floor there. The nook is that bay.

⭐ Width is RULED at 1 ft (owner, 2026-08-10). ⏸ Depth is open — the options are below, and this study does not pick one.

⭐ It is the part that has floor. The tree cutout — the hole with no floor in it — starts at X 12 and runs east to 60.5. Everything west of that line is deck: the west short-joist bay, 1'-0" of it, spanning from the west beam to the header. The nook is on the solid side of that boundary, which is exactly why it is usable at all.

⛔ Framed, not yet built. The four short joists J4–J7 (stations 54 · 64.5 · 75 · 85.5) and both headers are PLANNED — the as-built record says “no headers are hung; ran out of time”. And the west header carries its own open question: at 1'-0" of span against the east side’s 4'-0 1/4", “the west side is too short to need one” — delete or spec is OPEN (owner/PE). The LINE at X 12 is datum either way: it is where the cutout starts. Whether a member sits on it is not this study’s call.

Where the nook is, and what holds it up

North is up, east is right. Amber = the nook at its maximum depth. Hatched = the floor cutout, which has no floor. Tan = the oak. Heavy dark lines = the two cutout headers. Tan bands = the two double trimmers. Dashed gray = the short joists J4–J7, ⛔ not built.

A — CONTEXT PLAN. The headers frame into the trimmers; the trimmers sit 6" off the trunk’s N and S faces on both sides.

B — THE NOOK, ENLARGED. Width ruled; depth drawn as a band of options, none of them chosen.

How deep? — ⏸ not ruled

Depth is north–south, measured from the south chase wall on the S double trimmer, station 96, running north. North of wherever the nook wall lands, the space stays open, under the roof — the tree enclosure is a J, deliberately open north and northwest, and that openness is the reason the nook has a wall at all rather than simply petering out.

depth	nook wall at
1'-0"	station 84	the shallow option
1'-3"	station 81	—
1'-6"	station 78	—
2'-2 1/4" ⭐ MAX	station 69.75	lands on the trunk’s west midline — the midpoint of the two double trimmers, which the datum carries as `TRUNK_NS_CENTER_STATION`. The owner: “in theory can run to w trunk midline but we don’t need that much.”

⚠ The max is a derived ceiling, not a proposal. The pocket is described in prose — “west of the trunk to the west beam wall, north to about the trunk’s N–S midpoint” — and geometry.py deliberately declines to encode its extents, saying so in terms: no source states them as dimensions, and inventing one there is exactly what that section exists to stop. What the datum does carry is the midpoint itself, TRUNK_NS_CENTER_STATION 69.75 — and it is a midpoint twice over, of the trunk’s own N/S edges and of the two double trimmers, because the 6" standoff is held on both sides. That is what makes 2'-2 1/4" computable at all.

What fits in 1'-0"

⭐ The binding constraint is the standoff, and it is worth seeing plainly. An 18 in shelf — the depth the plan’s “1'-6" to the trunk” looks like it offers — overruns the west header by 6". That 6" is the trunk standoff. A shelf built to it would run out past the header, over the cutout, and finish at the bark. There is the whole argument for a 1 ft width, in one line.

Why a sway envelope exists at all — the tree moves. An oak in wind is not a post. The trunk leans, sways and comes back, all day, every day, and it does it hardest in exactly the weather you would be inside the cabin. Anything standing in the space a living trunk sweeps through does not get grazed; it gets pushed, repeatedly, by a tree. The build guide is explicit that this is sway, not growth: at roughly an inch of radius a decade, growth would take about sixty years to matter — “the driver is TREE SWAY, which is a this-winter event and is still unquantified” (2.6a §2b/§12). ⚠ Nobody has measured how far this oak actually moves, and observing it in a wind is owed before the engineer’s visit.

⚠ But that allowance is an EAST-side figure, and it is not what sets this nook. TRUNK_SWAY_ALLOWANCE 12 exists to hold the ridge and the east chase wall off the trunk: its own source says the 12 in “sits INSIDE the chase (ridge west face = trunk east face + 12)”, and the single assertion in the datum that consumes it checks the east chase wall — 54.5 + 12 = 66.5, closing exactly. Nothing in the datum applies it westward.

⚠ Two clearances live in the datum, they are not the same number, and they apply to different sides. TRUNK_STANDOFF 6 is the framing standoff, field-verified, held on every side of the trunk — it is what sets the cutout, the headers, the trimmers, and therefore the nook. TRUNK_SWAY_ALLOWANCE 12 is the east-side figure for the chase. Reading the 12 westward is the mistake this geometry invites, because both numbers are about the same tree — but only one of them is drawn on this side of it. On the west side the boundary is a straight line: the west double header at X 12.

The numbers, and where each one comes from

Quantity	in	what it is	datum constant
West beam CL ≈ west cabin wall line	X = 0	the nook’s west boundary and the study’s origin; the datum conflates the two, good to the ¾-in wall-span vs beam-c-c slop, and says so	frame note above `TRUNK_W_FACE_X`
Trunk east face — the anchor	54.5	TYPED. The tree does not move, so the whole E–W model is measured off it	`TRUNK_E_FACE_X`
Trunk diameter at deck height	36.5	DERIVED three independent ways. ⚠ FIELD VERIFY — one tape at deck height settles it	`TRUNK_DIA_DERIVED`
Trunk west face — the bark	18	= 54.5 − 36.5. This is the 18 in the plan reads — it is the distance to a tree, not to a surface you can build to	`TRUNK_W_FACE_X`
Trunk standoff	6	FIELD-VERIFIED 2026-07-24, both N and S; the datum is the trunk EDGE. It is what holds the framing off the bark on every side	`TRUNK_STANDOFF`
West double header	12	← the nook’s east boundary = 18 − 6. A straight line, framing into both double trimmers	`CUTOUT_HEADER_W_X`
West short-joist span	12	← the nook width. West beam CL → west header — the same 12.0, because it is the same bay	`SHORT_JOIST_SPAN_W`
Double trimmers	43.5 / 96	centerlines; 4 in overall each (± 2 in to a face). The headers frame into them	`TRIMMER_N_STATION` / `_S_` / `TRIM_PLY_HALF_W`
Trunk N and S edges	51.5 / 88.0	each exactly 6 in inside its trimmer’s inner face (45.5 / 94.0) — the symmetry that makes the midline computable	`TRUNK_N_EDGE_STATION` / `_S_`
Trunk N–S center station	69.75	= the midpoint of the double trimmers, (43.5 + 96) ÷ 2 — which is where the max depth lands	`TRUNK_NS_CENTER_STATION`
Max nook depth	26.25	= 96 − 69.75. ⏸ Not ruled — the pocket’s extent is prose-only and deliberately unencoded	derived here
Sway allowance	12	⚠ EAST SIDE ONLY. Its own source: the 12 in “sits INSIDE the chase (ridge west face = trunk east face + 12)”	`TRUNK_SWAY_ALLOWANCE`
East chase wall	66.5	= 54.5 + 12, closing exactly. This is the one and only place the datum spends the sway allowance	`CHASE_E_WALL_X`

⏸ OPEN. ✅ Width RULED at 1 ft (owner, 2026-08-10); ⏸ depth open. This study informs the interiors register’s items 2.2a (the west nook), 3.7 (sporadic shelves) and 3.10 (nook / hiding space), in docs/interiors/README.md. It does not edit the register — that is a different owner.

Deliberately NOT drawn — because no source in this repo states it, and this project does not guess:

Any shelf, cleat, ledger, bracket or fixing. The repo was searched and there is none: every cleat and ledger on file is an erection cleat, a ridge cleat or the loft ledger. How a shelf lands in this bay is unwritten, and the blocking map that would answer it (interiors 2.1) is itself open. The study draws the space, not the solution.
A nook depth. Four options are drawn because the owner named four; none is chosen here.
Any depth for the south chase nook. geometry.py gives none.
Wall sheathing, finish or thickness. The datum states a wall line, not which face of the wall it is. Whatever the finish is, it comes off the 1'-0" — the clear can only be less, never more.
Whether the west header gets built. Open (owner/PE). The line is datum; the member is not settled.

Sources

What	Where
Every geometric figure on this page	`treehouse/tools/draw/geometry.py`, DATUM v7 (2026-08-08c) — constants named in the table above
The 6 in standoff, field-verified both N and S, datum = the trunk EDGE	`geometry.TRUNK_STANDOFF` source cite — the as-built layout tab’s row “Trimmer offset from trunk EDGE”, MEASURED 2026-07-24
That the headers and short joists J4–J7 are PLANNED, not built; that the west header may not be needed	`geometry.SHORT_JOIST_COUNT` and `CUTOUT_HEADER_W_OPEN_NOTE` source cites — AB tab “no headers are hung; ran out of time”; d-brief §1 / LIVING memo §9
That the headers are doubled	`geometry.CUTOUT_HEADER_PLIES` — per the cut list (“the cut list says DOUBLE”); ⚠ recorded as genuinely uncertain, OWNER CONFIRM at build
The pocket runs west of the trunk to the west beam wall, north to about the trunk’s N–S midpoint — and that its extents are deliberately NOT encoded	`build-guide/phase2/2.12_loft_net_frame.md` §2, quoted in `geometry.py`’s enclosure-plan block beside the TODO that declines to encode it
That the enclosure is a J, open north and northwest; the south chase wall on the S trimmer/header junction	`geometry.py` enclosure-plan comment (Option E, owner ruling 2026-08-09)
The pocket adopted as the book/reading nook (no depth stated)	`build-guide/phase2/2.12_loft_net_frame.md` — “the NW pocket is ADOPTED as the book/reading nook… Extents still FIELD VERIFY”
Why the sway allowance is 12 in, that the driver is sway rather than growth, and that the amplitude is still unquantified	`build-guide/phase2/2.6a_ridge_support.md` §2b (the trunk EAST face + 12 in CLEAR stack) and §12
That the 12 in sits INSIDE the chase — i.e. that it is an east-side constant	`geometry.TRUNK_SWAY_ALLOWANCE` source cite; the only assertion consuming it is `rederive()`’s east chase-wall guard
That the tree must never be bound, chafed or loaded by the framing	`treehouse/MASTER-PLAN.md` — “do not caulk or bind the deck to the trunk — the tree must move freely”
Object sizes in the “what fits” figure	none — indicative, not specifications. Everyday objects at approximate size, so the width has a human scale

Four walls, four jobs

⏸ Partially ruled since freezing. The WEST wall (68.25 in) is full-height framed + sheathed and COUNTED — R7 (2026-08-13) killed 91–97 % of N–S torsion with it — and R39 (2026-08-16) closed the whole envelope circuit sheathed. Which further lines are COUNTED in Σb still goes to the shear-wall engine and the PE. Kept as the map of what each line is for.
⛔ FROZEN at datum v7 · 2026-08-10 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — NO WALL SELECTION IS IMPLIED — GENERATED FROM geometry.py, NOT HAND-DRAWN

Six lines, four different jobs — and the one that looks most like “the front” carries the least

A cabin wall is never just a wall. It is some mix of carrying roof, resisting racking, holding glass and being a way out — and here those jobs land unevenly. The north wall, the one you walk up to, does none of the structural ones at all. Read this before the shear-wall decisions; they only make sense once you can see which line does what.
★★ And one of the lines is not a fixed length. The west wall runs to the nook wall — so how deep you make a storage nook decides whether EAST or WEST is the longest N–S shear candidate. An interiors decision moves a structural line. That is the second thing this page is for.

⏸ RULING STATE: OPEN — this study informs shear-wall selection, it does not make it.
The wall lines and lengths are measured and settled. Which walls get sheathed is not. That selection goes to the 4_ENG_SW engine and is decided on numbers that do not exist yet. There are six lines here, two of them E–W, and every one of them is a candidate. If anything here reads as a selection, it is being read wrong.

1 · The six lines, in plan

Key

1	SOUTHstation 138	108.0 in	E–W
2	CHASE-SOUTHstation 96	54.5 in	E–W
3	EASTX 108	94.5 gross → 64.5 counted	N–S
4	CHASE-EASTX 66.5	52.5 in	N–S
5	WESTX 0 → the nook wall	54.0 – 68.25 in	N–S
N	NORTHdoor / porch gable	— free	—

Heavy black = a candidate wall line. Heavy black dashed, faded (west wall only) = the stretch of wall the unruled nook depth adds or takes away — the wall is continuous either way; only its north end is unsited. Gray dashed = the north wall, carried by nothing and carrying nothing. Red dashes just inboard = the wall sits over a drop beam and carries roof. Red hatched track held clear of the east wall = the ~30 in exit, drawn as a track the opening can sit anywhere along — its position in that wall is still open, so it is not drawn on the wall itself. Green dashed circle = the oak’s 12 in sway envelope around a 36.5 in trunk — FIELD VERIFY. Amber dashed = the ridge line at RIDGE_X 68.5.

★ The chase is an L, and it closes on the datum. Its south leg runs along station 96 from the nook edge (CUTOUT_HEADER_W_X 12.0) to its east leg (CHASE_E_WALL_X 66.5) = 54.5 in (4'-6 1/2"); taken all the way to the ridge centerline it would be 56.5. And the east leg’s 52.5 in is exactly 96 − 43.5 — the two double trimmers. Two legs, one L, both ends landing on members that already exist.

★★ The west wall does not stop at the south double trimmer. It is the nook’s west side as well as the cabin’s, so it runs continuously from the nook wall south to the S gable at 138: 42 + nook depth. And the matrix below calls it BEARING — it stands on the west drop beam. A bearing wall that stops mid-span is not a thing. Drawn solid over the 54.0 in every option has (nook wall at station 84), dashed north from there to station 69.75.

The west nook is the carve-out at the L’s west end and is why the south leg stops at X 12.0. Its width 12.0 in is ruled (it is the west short-joist bay, SHORT_JOIST_SPAN_W); its depth is open — the band is drawn to the stated maximum 26.25 in (station 69.75, the trunk’s west midline) with its north edge dashed, and the shallow option’s wall (station 84) dotted inside it. That is an envelope of the options, not a chosen line — and it is the figure that sets the west wall’s length.

2 · How long is each line — once you only count what is full height

Segmented shear sums full-height segments only. An opening does not weaken a wall a little; it removes its own width from that wall’s counted length (Σb). So the east wall’s ~30 in exit takes it from 94.5 in down to 64.5 in of counted line — a bite out of the bar, not a symbol on a plan.

⚠ And one bar is a range. The west wall’s length is 42 + the nook’s depth, and that depth is ⏸ not ruled — so WEST is drawn as a band from 54.0 in to 68.25 in, with the superseded 42.0 in figure struck where it used to end. No depth is picked here.

★★ It is a crossover, not a winner — and the thing that moves it is a storage decision. EAST counted is 64.5 in. WEST is 54.0 in at a 12 in nook, 64.5 in at 22.5 in — a dead tie — and 68.25 in at the 26.25 in maximum, where WEST is the longest N–S line in the building. The page used to claim a flat ratio here. There is no flat ratio; there is a 22.5 in crossover.

★★ An interiors storage decision moves a structural candidate line. How deep the book nook goes decides whether EAST or WEST is the longest N–S shear candidate. That makes the nook’s depth not a free interiors preference — it is the sharpest case in this project of a finish choice spending, or buying, structure. Whoever rules it has to be told. This is exactly what a wall study is for.

★ What survives from the old headline, and is stronger for being conditional: an opening still does not disqualify a wall. East holed stays competitive with west whole across the range — ahead of it below 22.5 in of nook, level at it, and within 5% of it even at the maximum. Picking the shorter line as the sole N–S shear wall because the longer one has a door in it would still have been the actual mistake. What is gone is the idea that the comparison has one fixed answer.

nook depth (2.2b)	nook wall at station	WEST = 42 + depth	EAST counted	EAST / WEST
12 in	84	54.00 in	64.5 in	1.19×	EAST is the longest N–S line
15 in	81	57.00 in	64.5 in	1.13×	EAST is the longest N–S line
18 in	78	60.00 in	64.5 in	1.07×	EAST is the longest N–S line
22.5 in ★ the crossover	73.5	64.50 in	64.5 in	1.00×	★ dead tie
26.25 in (the stated MAX — trunk W midline)	69.75	68.25 in	64.5 in	0.95×	WEST is the longest N–S line

⏸ No depth is chosen here — the table is the whole open range, register row 2.2b. The crossover is not typed: it falls out of 64.5 − 42, and build_study.py::_check() re-solves it at generation.

Read the bar as an accounting of length, not as a plan. The bite is drawn at the end of the east bar because Σb does not depend on where the opening sits: centered or offset, the counted total is the same 64.5 in. The position is a live variable and is deliberately not drawn.
And the two sums are separate. The three N–S lines add up with each other; the two E–W lines add up with each other; neither group helps the other.
⚠ Two bars are drawn open — chase-south AND west — because no source states either wall’s height or its full-height continuity, and a wall that stops short of the plate contributes nothing to Σb. 54.5 in and 54.0–68.25 in of line are facts; the same figures as Σb are not. West being a bearing wall makes full height likely — and likely is not stated, so this page does not invent it.

3 · Wall × job

wall	RESIST RACKING (shear)	CARRY ROOF (bearing)	HOLD GLASS	BE A WAY OUT
1SOUTHstation 138 · 108.0 in	▲ CANDIDATE — E–W the longer of the two E–W lines	○ no — gable end the ridge lands on posts, not on this wall	● gable peak only above the plate — costs no Σb	○ —
2CHASE-SOUTHstation 96 · 54.5 in	▲ CANDIDATE — E–W ★ the second E–W line — so this direction is not a single-wall system. ⚠ height not stated: line length, not credited Σb	○ no — no roof on it but it stands on the S double trimmer, a real Year-1 member	○ not ruled	○ — interior line: the loft’s north boundary, and a frame attachment point (owner)
3EASTX 108 · 94.5 gross / 64.5 counted	▲ CANDIDATE — N–S longest N–S line even after the bite	● YES — over the east beam	○ not ruled	● the ~30 in exit position within the wall OPEN
4CHASE-EASTX 66.5 · 52.5 in	▲ CANDIDATE — N–S ★ a line the guide’s original scheme never counted — and the shortest N–S line of the three	○ no — mid-span, no beam under it	● plexi — it is the chase wall	○ —
5WESTX 0 · 54.0–68.25 in	▲ CANDIDATE — N–S ★★ its length is not fixed: 42 + nook depth (2.2b). Longest N–S line above a 22.5 in nook, shorter than EAST below it. ⚠ height not stated: line length, not credited Σb	● YES — over the west beam which is why it cannot stop at the S trimmer	○ not ruled	○ —
NNORTHdoor / porch gable · no length credited	○ none — STRUCTURALLY FREE	○ no	● gable peak (porch gable)	● the door

▲ = a candidate line in that direction — never a selection. ● = the wall does this job. ○ = it does not, or it is not ruled. Bearing walls are the two outer N–S lines, because those are the two that stand over the drop beams.

Two frames live under this cabin and this page uses the WALL one. WALL_SPAN 108.0 in (9'-0") is a wall-line-to-wall-line figure; the as-driven beam centerlines are S_CROSS_CC 108.75 and N_CROSS_CC 108.375 — different numbers for the same distance, and the frame is not square. Nothing may be fabricated to a figure on this page.

4 · The gable that has a post in it

Σb is measured in the full-height wall — floor to plate, PLATE_Z 66 in (5'-6", LOCKED). Above that line the gable is a triangle that carries no roof: the ridge lands on its four posts, not on the gable wall. So a gable-peak opening still costs zero Σb — that finding is unchanged.

⚠ What is not unchanged is the shape of what is possible up there. One of those four posts — the S gable ridge post — stands in this wall plane, X 66.5 → 70.5 (RIDGE_X 68.5 ∓ POST_W/2), running floor to ridge soffit. There is no full-width opening to be had; there are two openings, one each side of it.

★★ Near the post, symmetry is nearly free. At the owner’s 12 in working width the two sides differ by 11% in available area. Run them out to the eaves and they are 47% apart — and the east one has tapered to nothing at the wall line. The asymmetric roof only bites as you move outboard.

The width sweep — each side of the post

width each side	west zone, X	east zone, X	west area, in²	east area, in²	W over E	height at outer east edge
8 in	58.5 → 66.5	70.5 → 78.5	394	366	8%	40.3 in
12 in ★ working figure	54.5 → 66.5	70.5 → 82.5	572	517	11%	34.9 in
20 in	46.5 → 66.5	70.5 → 90.5	891	752	18%	23.9 in
37.5 in (to the eaves)	29.0 → 66.5	70.5 → 108.0	1,412	961	47%	0.0 in

★ 12 in is the owner’s working figure, not a ruling (“12 in symmetric each side or whatever our tolerance is”). Symmetric about the post and not run to the eaves are both explicit; the widths are a sweep, and the sizes are still ⏸ open.

⚠ These are the triangle’s heights, not window heights. Every height and area above is the gable triangle’s own at that station — the space the construction allows. No opening height is drawn or implied, and the real clear space is lower still: the rafter soffit planes pass through the birdsmouth seats, 1.97 in below this line on the west slope and 2.05 in on the east.

The triangle is the datum’s slope-defining construction: PLATE_Z at both wall lines, rising APEX_RISE 54 in over runs 68.5 / 39.5 to RIDGE_X — west slope 38.25°, east 53.82°. ⚠ APEX_X 69.536 is a different point (where the two rafter top planes meet) and is not this triangle’s apex; heights taken off it run about an inch high. The dashed member on the post is the ridge beam end-on, 4 × 10 — its top sits 0.37 in above the construction line, because the construction line is a slope datum, not the roof.

Only the north and south walls have a triangle — east and west run under the eaves at constant plate height. The north gable has the same condition, a ridge post at station 16 in its plane; this study draws the south.

Deliberately NOT drawn

Which walls are sheathed, and which line is “the” shear wall. OPEN — handed to 4_ENG_SW, decided on numbers that do not exist yet.
The exit opening’s position. A live variable. Drawn as a band and as a bite; never as a located object.
Chase-south’s HEIGHT and west’s HEIGHT, and therefore either one’s Σb credit. Both lines are derived; neither height, nor either wall’s full-height continuity, is stated by any source.
The nook’s depth — and therefore the west wall’s exact length. Drawn as the whole open range and as a crossover; never as a chosen number. Nor anything west of X 12.0 at station 96 beyond the ruled 12.0 in width.
Gable-opening HEIGHTS, heads and sills. The pair is drawn as two zones of the triangle at a swept width; nothing on this page is a window.
Header sizes, stud counts, nailing, hold-downs, sheathing type. None ruled.
Segment aspect ratios and per-segment capacity. That is the engine’s work, on inputs that are not in hand.
Glazing anywhere except the chase east wall (which is the plexi wall) and the gable peaks.

Built from geometry.py at DATUM v7 (2026-08-08c), plus the interiors register 2026-08-10 · calcs v3 workbook spec §2 (rows 1.2a, 1.2c, 1.4a, 2.2a-rev, 2.2b) for the wall inventory, the ~30 in east opening and the gable-pair framing. Datum-derived and register lengths are cross-asserted at generation (build_study.py::_check()); a disagreement stops the page rather than getting drawn. • The west wall is drawn as a range (54.0–68.25 in) because row 2.2b is ⏸ open; the superseded 42.0 in figure is shown struck, and _check_text() refuses to write a page that states it unretired, or that quotes the flat east-over-west ratio it produced. • FROZEN 2026-08-10. Do not regenerate. A study records what was known when a question was framed; re-running it against a later datum produces a different document and destroys that record. • This page is never to be measured from.

Ladder siting — the WEST edge, or the NORTH edge?

⭐ Ruled 2026-08-12: WEST is the default — the near-trunk arrival, the whole point of climbing up beside the oak. The tail taper (ruled the same day) lifts the landing to 4'-11 5/16"; NORTH stays documented as the alternative; a Year-3+ self-supported stair is reserved, not ruled out.
⛔ FROZEN at datum v8 · 2026-08-12 · not regenerated — it records what was known when the decision was made

STUDY — NOT PART OF THE DESIGN — BOTH OPTIONS ARE CANDIDATES, NEITHER IS CHOSEN — GENERATED FROM geometry.py, NOT HAND-DRAWN

Where does the ladder land — the WEST edge, or the NORTH edge?

One climb, two places to put it. The difference is not taste: on the west you top out under the rafter tails, and on the north you top out under open sky. This page draws both to the datum and decides neither.

⭐ REVISED 2026-08-12 — OWNER RULING: EAVE TAILS TAPERED TO A 2.0 IN TIP (DATUM V8); THE PRE-TAPER V7 FIGURES ARE KEPT BELOW AS THE RECORD THE RULING WAS MADE AGAINST.

This page was drawn on datum v7 (2026-08-08c), the owner read it, and ruled the tails tapered on their underside to a 2 in tip — which is now datum v8 (2026-08-12). The ruling was made on the figures below, so those figures are kept, dated and labelled rather than overwritten: the west section draws the adopted cut solid and the superseded v7 soffit as a dashed ghost. ⚠ The siting question itself is UNCHANGED and still open. The taper moved one number on one side; it did not choose an edge.

⚠ This is not a license to re-run a study. It is legitimate here for a narrow reason: the page had never been published, and the ruling it triggered is about the very geometry it draws. A study whose decision has landed is frozen — re-deriving it on a later datum destroys the record of what was known.

⏸ RULING STATE — OPEN. Nothing here is selected. Interiors register 0.4b (“ladder siting buys headroom at 9.46 in per foot east”) is a rule, not a position, and it is still open; 3.2 (the door) is open; 0.4a is an unresolved discrepancy in the datum itself (below). A north landing would supersede a standing ruling — also below. All of that is an owner call.

⚠ WHAT IS DECIDED ALREADY, and is NOT reopened here. The mounting is decided — a lean-to ladder, no deck penetration (access memo 2026-07-04 §4). The head bears on the SOLID RIM, not on the joist-tip lip (same memo, 2026-07-22). A dedicated grab rail at the ladder head is ruled, with the cabin corner as the secondary handhold. And 0.3 is the standing filter on all of it: kids up alone from age 4, permanent ladder, no lock. ⚠ No project source rules the ladder TYPE or its ANGLE, and this page does not set either.

⚠ Before anything else: the datum contradicts itself about where the ladder lands

Register row 0.4a, raised 2026-08-10, still OPEN. Twelve lines apart inside one comment block, geometry.py says both of these:

“The ladder lands in the unencased NW quadrant and you step north into the porch.” — a WEST-edge landing, in the zone between station 43.5 and the trunk’s N–S midpoint at station 69.75.
“the NW porch is where the ladder lands.” — the porch is NORTH of station 43.5, which is a different place.

The owner said 2026-08-10 “on outside of structure on ladder”, which narrows it but does not settle which edge. So “the standing plan” is not a single place, and this page draws the west reading as the west candidate rather than pretending the datum speaks with one voice. ⚠ When it is ruled, both lines get fixed in one commit — that is the discrepancy protocol, and it is not this page’s job.

① The plan — both candidates, and the roof over them

How to read it. The roof is drawn semi-transparent over everything, with its rafters at 16 in o.c. showing the rhythm only. The roof’s north edge is the gable line at station 16 — this design has no rake overhang (2.6a §3a.3), so north of that line there is nothing overhead at all. The west candidate is a zone, not a point: the NW pocket’s extents are prose in the datum and deliberately not encoded (geometry.py’s own TODO says so, in terms), so it is drawn between the two lines the prose does name. The north candidate is a zone for a different reason: no position along the rim has been chosen, and the headroom over the walk south changes with where you put it.

The ceiling gains 9.46 in for every foot you walk east. 66 in at the west wall line · 72 in at X 7.61 · 75.46 in at X 12 · capped at the ridge-beam soffit 107.635 in. That is register 0.4b’s rule, re-derived here from APEX_RISE / RIDGE_X — and it is exactly why where the ladder lands is a headroom question and not a preference.

② The two arrivals, at one vertical scale

⚠ The west collision, stated as arithmetic — and what the taper bought. The west deck edge sits 9.625 in outside the wall line (JOIST_OVERHANG_EW 5.625 + SIDE_RIM_W 4, closing on DECK_OVER_RIMS_EW 128), and the eave edge, 15 in NOMINAL past the wall line, reaches 5.375 in past the landing itself — so a climber on a rim-leaned ladder rises directly under the rafter tails, and that has not changed.

⭐ What changed is how much room is under them. The taper lifts the underside by the ledge’s share of the tip gain — (9.625 / 15) × 4.457 = 2.860 in over the landing, and the whole 4.457 in out at the eave edge:

over the landing: 56.44 → 59.30 in (4'-8 7/16" → 4'-11 5/16") — the upright-stature threshold at the moment of arrival
at the eave edge: 52.20 → 56.66 in
the duck: carrying register 0.4’s 72 in stature up the 4:1 line, the head now meets the cut tail at X -13.24 with the feet 14.5 in below deck level — it was 19.4 in, so the duck happens 4.9 in later up the climb (and 16.9 in on the slope-defining envelope, which is the figure register 0.4’s plane implies and the one this page led with at v7)

⚠ It is a real gain and it is not a fix. A 72 in adult still ducks 12.7 in at the landing (it was 15.6), still under the tails, still at the least stable instant of a climb that register 0.3 says will be made by a four-year-old, alone, sometimes in the dark.

⭐ The published plane, the real soffit, and the cut — three lines, and which one governs where. Every arrival figure in register 0.4 / 0.4b is measured on the datum’s slope-defining plane, which meets PLATE_Z exactly at the wall line. The rafter’s own soffit — geometry.py’s _soffit_w — passes through the birdsmouth seat, EAVE_SEAT_W 2.5 in inboard, and therefore sits 1.971 in lower everywhere. That correction is unchanged and still runs against the west option inside the room: a 72 in adult stands upright at X 10.11, not X 7.61, and the taper does not move that by a thousandth — datum v8 rule C keeps the cut wholly outboard of the bearings, so nothing indoors moves at all.

⚠ But OUTBOARD of the wall line the order of the lines now flips. Over the landing the cut tail sits at 59.30 against the envelope’s 58.41 — 0.889 in ABOVE the published plane, because the 2.860 in of taper outruns the 1.971 in correction. So register 0.4b’s 4'-10" is now conservative at the deck edge, where it used to be optimistic. ⚠ Only there. At the wall line and everywhere inside it, 0.4 and 0.4b still read high by 1.971 in.

⚠ The north price, and this page does not pay it. The north rim is the cantilever tip: 18 in NOMINAL past the post line, which is the PE cap (“18 exactly = the PE cap; if field asks >18, shift the rim south”), with the cross centerline 2 in inboard of that and the NE tip field-locked at 17.25. A ladder-head reaction at that tip is checked in no document on file — not in the calcs workbook, not in any memo. It is the same objection the 2026-07-04 memo raised against its own option B (“people congregating + any stair reaction at the tip is a live-load item”), and it has never been answered. This study states it and refuses to clear it. It is a question for the engineer, and it is the north option’s real cost.

③ What you climb toward, from the meadow

⚠ This one is mirrored, and deliberately: an exterior elevation is taken from outside, so facing south puts east on your left and reverses the plan above. The generator derives that from the view’s own label and re-checks the drawn mapper against it, because a stored flag and a printed label are two facts that can disagree. The roof here is drawn as the ceiling you walk under — the ridge beam hangs 10 in into the room, so the high point of this figure is the soffit at 107.635 in, never the seat-line construction point above it.

West against north, honestly

	● WEST edge — the standing plan (as far as it is one)	● NORTH edge — proposed
where the head tops out	the west rim, in the zone stations 43.5–69.75. ⚠ a ZONE: the pocket’s extents are prose, not datum values.	the north rim, station 0. ⚠ a ZONE too: the position along the rim is not chosen.
what is overhead	the rafter tails, now cut back. 59.30 in to the tapered tail as ruled (v8) — was 56.44, and 58.41 on the envelope. The tails still run 5.375 in past the landing; the taper changes the room under them, not the fact of them.	open sky — 16 in of it, because the roof stops at the gable line and there is no rake overhang. Unaffected by the taper.
the climb-over moment (0.3)	the head meets the cut tail with the feet 14.5 in below the deck — 4.9 in later up the climb than the 19.4 in the pre-taper soffit gave. Still a duck, and still at the worst instant.	nothing overhead at the transition at all. The hands, the gate and the grab rail are the whole story.
standing upright	not until X 10.11 (real soffit) — 19.74 in inside the wall line from where you landed. ⚠ The taper buys none of this: it is a cut outboard of the plate, and indoors nothing moved.	immediately, and it stays that way: at the door station the porch soffit is 92.32 in (94.37 envelope) above the deck.
arrival experience	⭐ the tree-facing arrival — you come up beside the oak, through the open NW corner. The 2026-07-04 memo calls it “the whole charm” and it was the owner’s original vision.	frontal. You top out facing the door across the covered porch — more like arriving at a house, less like arriving at a tree.
path to the door	land in the NW quadrant → step north into the porch → turn east to the door at X 66.5–108. Three moves.	straight south: 43.5 in to the door line, of which 27.5 in is roofed. One move.
weather	NW is the site’s lake-effect corner (2026-07-04 §3) and the climb is under a dripping eave edge that overhangs the landing. Ice forms where you grab.	also a weather face, and the top-out is the uncovered cantilever — no roof over the last rung, 16 in of open sky, then cover.
guard interaction	a gap in the west return run; a self-closing, self-latching gate is already ruled wherever the climb lands (2026-07-04 §6).	a gap in the north run — and the four Year-1 guard posts have their bases embedded in that rim (1.7 §6a). ⚠ No source states their E–W stations, so post-versus-opening cannot yet be checked either way.
structure	the head bears on solid rim over the west beam line — ruled, and the rim is backed. ⚠ The west beam’s own mid support is the unresolved west-mid TAB (S-104, PE + arborist gated); its planned station is quoted in a different (up-string) frame and is not converted here.	⚠ OPEN. The head bears on the tip of the 18 in cantilever and nothing on file checks that reaction.
standing rulings touched	consistent with D3 — the zone lies wholly south of the gable line. ⚠ But see 0.4a: the datum states the plan two ways.	⛔ SUPERSEDES D3 (2026-08-03: “Ladder head lands SOUTH of the gable line”). Station 0 is north of station 16. That needs a dated owner supersession, and eventually a datum comment pass — never a silent change.

⛔ What choosing NORTH actually costs in paperwork

D3 is a live ruling and a north landing overturns it. The 2026-08-03 memo adopted it flatly: “D3 — Ladder head lands SOUTH of the gable line. Free, layout-only, and it keeps the ladder→door path clear of anything at the gable plane. Adopted regardless of D1.” The memo is archived, and archived decisions still govern where they were not superseded — D3 was not.

⭐ But its stated reason was written for a door that no longer exists. D3’s justification is the ladder→door path, and in that memo the door was D2’s west-facing door on the N–S connector wall. D2 is DEMOTED — the owner adopted Option E on 2026-08-09 and the door is now north-facing at station 43.5, X 66.5–108. Whether D3’s reason survives the door move is exactly the owner question this page exists to put — and it is a question, not an answer: a rejection is only as durable as the fact under it, and reopening one is done with a new dated row, never by deleting the old one.

⚠ Three published figures this page could not reproduce

Drawing a thing to the datum is a check on the prose about it. These came out of that check. ⛔ NOTHING HAS BEEN CORRECTED IN ANY SOURCE FROM HERE — a study reports the arithmetic, the owner rules, and the owning document is fixed in its own commit. All three are raised, none is resolved.

1 · The west and east ledges look swapped in register 0.1b

The register’s “standing room” table gives WEST 10.4 in and EAST 9.6 in. Derived from the datum: the deck runs JOIST_OVERHANG_EW 5.625 + SIDE_RIM_W 4 = 9.625 in outside each beam centerline (and 108.75 + 2 × 9.625 = 128, the published over-rims width). The wall line and the west beam centerline are the same line in geometry.py’s own E–W frame, so the WEST ledge is 9.625; the 0.75 in of wall-versus-beam slop falls at the EAST, making that one 10.375. The register’s pair is the right pair with the sides exchanged. ⭐ It does not change this page’s argument — 9.625 is the west figure in either reading of the two E–W frames, and it makes the west arrival taller, not shorter.

2 · Register 0.4b’s “4'-10” at the west deck edge survives, but not by the route it was computed

On the derived 9.625 in ledge the envelope gives 58.412 in = 4'-10 7/16", which is 4'-10" to the nearest inch — the published figure. On the register’s 10.4 it is 57.821, which also rounds to 4'-10". The headline holds either way; the 0.75 in it was computed from does not.

3 · Register 0.4’s “a 6 ft adult clears at X 7.6” is an ENVELOPE figure

Against the datum’s slope-defining plane it reproduces exactly: 66 + 7.611 × 0.788321 = 72.00. Against the rafter’s actual soffit, which passes through the birdsmouth seat 2.5 in inboard, the same adult clears at X 10.111 — 2.5 in further into the room. Row 0.4 is marked DECIDED (geometry), so the difference is worth an owner’s eye. It is the same 1.971 in offset throughout, and it always runs the same way.

• And one thing that reconciles, stated so nobody re-derives it

NW_PORCH_DEPTH_NOMINAL is 47 in (“the deep (~47 in) NW porch”, D2 memo §1) while the as-built chain puts the north rim 43.5 in from the cabin’s north wall line. The difference is 3.5 in — exactly WALL_STUD_THK, which would follow if the ~47 were measured to the far face of that wall. ⚠ No source says that, the constant is NOMINAL and its memo is superseded on the layout; treat it as FIELD VERIFY, not as a resolution.

Sources, in priority order

1 · Building code / regulation. The drawn pitch is [REG] 29 CFR 1926.1053(b)(5)(i), which states it on the working length — “the horizontal distance from the top support to the foot … is approximately one-quarter of the working length”, i.e. 75.52° and a foot 21.43 in out. The trade rounds that to “4:1”, which is rise-to-run and gives 75.96° with the foot 20.75 in out — a difference of 0.68 in on the ground. ⚠ That is an OSHA construction-workplace rule, cited as the established practice for an illustrative angle. No project source rules a ladder angle, and this page does not set one. Separately: a ladder is not a code-compliant primary means of egress for habitable space (2026-07-04 §4) — an AHJ question, unanswered.
2 · Definitive sources — this project’s own decided record. geometry.py at DATUM v8 (2026-08-12) for every dimension; the 2026-07-04 access memo for the lean-to mounting, the solid-rim bearing, the grab rail and the A/B/C option table; the 2026-08-03 north-gable memo for D3; the 2026-08-09 porch-options study for Option E; the interiors register for 0.3 / 0.4 / 0.4a / 0.4b / 3.2.
3 · Trade practice. Not cited, and not needed — nothing on this page details a joint, a fixing or an assembly. When a ladder is actually specified, that is where its head fixing and its gate hardware will have to come from, after the two sources above.

Deliberately NOT drawn

A choice. Neither option is selected and no recommendation is made.
Any position along the north rim. The north candidate is a sliding zone; where it goes interacts with four guard posts whose stations no source states.
The ladder itself — type, width, rung spacing, tread form, material, how far the rails run above the landing, and how the head is fixed. The mounting is decided; nothing else about the ladder is, and the dashed outline is a study outline at an illustrative pitch with its arithmetic printed beside it.
The gate, its hardware, its hinge side and its swing. A self-closing, self-latching gate is ruled; nothing about it is specified, here or anywhere.
The grab rail beyond noting that it is ruled — no height, no length, no fixing.
Any structural clearance of the north cantilever tip under a ladder-head reaction. That is the one thing this page most wanted to settle and most refuses to. It goes to the engineer.
Any child stature figure. Register 0.3’s filter is stated in words. The 72 in figure on both sections is register 0.4’s own 6 ft adult, carried up the line as an illustration.
The NW pocket’s extents as dimensions. geometry.py refuses to encode them and so does this page.
The guard posts, on either edge — no source gives their E–W stations.
The roof’s finished outline, its fascia, soffit boards or drip edge. The sections draw the structural soffit plane; a finished soffit would sit at or below it, and would make the west case worse, never better.
The west-mid support. Open (S-104, PE + arborist), and its station is published in a frame this page does not convert.
Anything at all about stairs. The 2026-07-04 memo keeps a Year-3 stair on the table; that is a different question with a different footing.

⭐ What the taper did and did not do

	v7 — pre-taper (the record)	v8 — as ruled 2026-08-12	change
underside over the WEST landing (X -9.625)	56.442 in · 4'-8 7/16"	59.301 in · 4'-11 5/16"	+2.860 in
underside at the eave edge (X -15, NOMINAL)	52.204 in	56.661 in	+4.457 in (the whole tip gain)
feet below deck when the head meets it	19.38 in	14.48 in	4.90 in later up the climb
residual duck for register 0.4’s 72 in adult	15.56 in	12.70 in	−2.86 in
upright station INSIDE the room	X 10.111	X 10.111	nothing — the cut is outboard of the bearings
tails past the landing	5.375 in	5.375 in	unchanged — the overhang did not move
the NORTH option, in every respect	no roof over the top-out either way		unchanged

⚠ The tip carries an open engineering item. Datum v8 rule D: [PE] TO CONFIRM — eave snow/ice on the 2 in cantilevered tip. That check is not on file, this page does not make it, and the figures above are geometry, not a capacity statement. A tapered or plumb-cut rafter tail is established framing practice; no source on file describes this project’s tail cut, and none is invented here. The construction set carries the same flag on 2/S-301 and 3/S-301, with its references ladder.

Built from geometry.py at DATUM v8 (2026-08-12) — PLATE_Z, RIDGE_X, APEX_RISE, both roof planes and both rafter SOFFIT planes, EAVE_OVERHANG_NOMINAL, Z_RIDGE_BOT, the AS-BUILT N→S station chain, the trunk band, the E–W deck build-up, DECK_ABOVE_GRADE_NOMINAL, N_CANTILEVER_NOMINAL and the Option E enclosure prose — plus the interiors register 2026-08-10 (rows 0.1b, 0.3, 0.4, 0.4a, 0.4b, 1.2e, 2.1, 3.2) and the two dated access memos for the figures the datum does not own. Every drawn number is derived at generation; the generator asserts the slope triangle closes on PLATE_Z at 0.0000 at both wall lines, that the collision and the D3 relationship are derived rather than claimed, and it scans the composed page for forbidden claims before writing it — any of them failing stops the build rather than shipping a drawn guess. • REVISED AND RE-FROZEN 2026-08-12 (datum v7 (2026-08-08c) → v8 (2026-08-12), owner-instructed, before publication; every superseded figure kept and labelled). FROZEN 2026-08-12. Do not regenerate. A study records what was known when a question was framed; re-running it against a later datum produces a different document and destroys that record. • This page is never to be measured from.

Ruled — archived. Decisively decided; the full frozen studies are embedded below, collapsed. (Owner curation 2026-08-13, extended 2026-08-16 under the same standing rule.)

📁 archived · porch-options — click to view

⭐ RULED Option E (the larger, full-width porch), 2026-08-09 — same day it was drawn. The unadopted options live only in here.
⛔ FROZEN at its own datum · 2026-08-09 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-09-porch-options/

STUDY — NOT PART OF THE DESIGN — NEITHER SCHEME IS ADOPTED — GENERATED FROM geometry.py, NOT HAND-DRAWN

Two porch options, drawn to the same datum

Solid = Year-1 as-built, from the datum. Dashed black = Year-2 wall lines — prose-only, NOMINAL, never dimensioned. Amber squares ①②③④ = the four locked ridge posts. Green dashed circle = the trunk's 12 in sway envelope. Yellow fill = porch.

E — LARGER PORCH

Full width. Cabin north wall moves south to the N double trimmer (43.5). The NE “hallway” becomes porch. Entry N→S, hinged.

D — SMALLER PORCH (current / D2)

Porch stops at the chase line (66.5). The NE leg stays cabin. Entry W→E through the connector, sliding barn door.

What actually changes. Only walls move, and walls are prose-only NOMINAL. Stations 16 / 43.5 / 96 / 138 are identical in both, so the spans stay 27.5 / 52.5 / 42.0, the reactions stay 127.9 / 806.1 / 959.9 / 271.6 lb, and the four-post lock stands. This is a layout question, not a re-lock.

	E — larger	D — smaller
Porch width	128 in (full)	77 in
Porch depth to cabin	rim → sta 43.5	rim → sta 43.5 (west) / 16 (east)
Cabin lost to porch	~8 ft² (the NE hallway)	—
Entry	hinged, out-swing onto porch	sliding barn door
Barn door possible	no (~15 in of park)	yes — the point of D
Post ① (N gable)	stands free in the porch	inside the north wall
Post ② (N mid)	inside the new north wall	west face flush on the connector line — zero clearance
Shape to sheathe/brace	rectangle	J

Entry section — option E, looking north at the door wall

The wall runs X 66.5 → 108. The apex sits inside it at APEX_X 69.536, so both roof planes appear — the superseded arithmetic used one straight line and a wrong anchor, and made a 6'-8″ door look impossible.

Door head	East jamb reaches	Framing available	Leaf
80″ (6'-8″)	X 98.8	32.3 in	~24 in
72″ (6'-0″)	X 104.6	38.1 in	~30 in
66″ (5'-6″)	X 108.0	41.5 in	~34 in

Ceiling at the east wall is 67.42 in, not PLATE_Z 66 — the rafter sits above its seat. Under the connector wall (option D) the binding head is the ridge underside, Z_RIDGE_BOT 107.6 in, which is where D’s 84″ barn panel fits and E’s door wall cannot reach.

📁 archived · wall-elevations — click to view

⭐ P·PLINTH ruled the plexi default 2026-08-12; superseded as the openings reference by the two 2026-08-13 opening studies. Its 61.5-in clear height is v7-era — correct as frozen (the live figure is 60.5 at v9).
⛔ FROZEN at its own datum · 2026-08-10 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-10-wall-elevations/

STUDY — NOT PART OF THE DESIGN — CANDIDATE OPENINGS ONLY, NOTHING HERE IS SELECTED — GENERATED FROM geometry.py, NOT HAND-DRAWN

Six walls, and each one can hold glass in a different way — or not at all

The plan study says what each wall is. This one asks what can go IN it: where a window can sit, how high its sill must be, which wall can be glass wholesale, and where a vent belongs. Everything is drawn to one common scale so the six can be read against each other — and the first thing that comparison shows is that they are not all the same height. The ceiling in this room is not flat, and two of the six walls are far taller than the rest.

★★ REVISED 2026-08-10, after owner review, before publication. The first draft of this page had two defects.
1 — it treated every wall as the same clear height. That figure is an eave-wall figure. geometry.py’s HEADROOM block (2.12 §9.2, ruled 2026-08-06) says it outright: “at the E/W eave walls the ceiling IS the plate line — the roof rises from there”. Corrected in §1, and the generator now refuses to write the eave figure as either interior wall’s clear height.
2 — it applied the kid-safe sill band to the fixed plexi panel — that is, to the very thing register 0.3 proposed as the remedy for breakable glass. Owner ruled the same day (register 0.3a): the band governs glazing that can OPEN or BREAK — glass and operable sash — and does not govern fixed plexi. Re-scoped throughout.

⏸ RULING STATE: OPEN — this study makes the window decisions DECIDABLE. It does not make them.
Register 1.2 (siting), 1.2c (the gable pair — 12 and 18 in are working figures), 1.3 / 1.3a / 1.3b (the plexi wall and its product) and 3.3a (venting) all remain open. 1.2a is decided — the second exit is on the EAST wall — but the egress-capable window’s siting is still open, and this page draws candidates, never a choice.

1 · The ceiling is not flat — and that governs everything below

The roof is a ridge-beam gable running north–south. So the ceiling over any east–west line rises from the plate at the west wall to the ridge, and falls again to the plate at the east wall. Two consequences the first draft missed:

At the eave walls the ceiling is the plate line — the roof rises away from them. Their height is the wall height, and nothing more.
Under the ridge, headroom is measured to the beam SOFFIT, Z_RIDGE_BOT 107.635 in, “because the 10 in ridge hangs into the space” (2.12 §9.2). It is not measured to Z_APEX (120 in), which is a seat-line construction point: nobody stands under it, and no view may print it as a head-height figure. This page obeys that literally, and its generator asserts it.

★ You arrive at the ladder with 66 in over your head, and you gain 9.46 in for every foot you walk east. A 6 ft adult is stooping at the west wall line and upright by X 7.6 — about eight inches in. That single number (0.7883 in per inch, the west plane’s own slope) explains every wall height on this page. Register row 0.4 records it.

The ceiling, station by station

station	ceiling above the floor		what is there
X 0	66.00 in	5'-6"	the WEST eave wall — the ceiling IS the plate
X 7.611	72.00 in	6'-0"	a 6 ft adult stands upright from here
X 12	75.46 in	6'-3 7/16"	the nook edge, and CHASE-SOUTH’s west end
X 18	80.19 in	6'-8 3/16"	the trunk’s west face
X 36.25	94.58 in	7'-10 9/16"	the trunk centerline
X 52.81	107.64 in	8'-11 5/8"	level with the ridge-beam soffit
X 66.5	118.42 in	9'-10 7/16"	the beam’s west face — the CHASE-EAST line

⚠ These are the envelope, not a clearance. The plane drawn is the datum’s slope-defining construction, which meets the plate at exactly 0.0000 at both wall lines; the real rafter soffit sits a little lower, because it passes through the birdsmouth seats inboard of them. At the ridge the beam interrupts the plane altogether: it hangs 10.79 in below it — the depth of the beam, and the reason the soffit is the datum.

★★ The clear height, wall by wall

wall	the ceiling above its line	floor → ceiling	clear
WESTX 0	the plate line — the roof rises away, east	66.0	61.5
EASTX 108	the plate line — the roof rises away, west	66.0	61.5
SOUTHstation 138	the gable triangle — on the plate at both wall lines	66.0 to the plate	61.5 in the wall	plus the triangle above it
NORTHstation 43.5	the same triangle, east half only	66.0 to the plate	61.5 in the wall	plus one slice above it
CHASE-EASTX 66.5	★★ THE RIDGE-BEAM SOFFIT — level, the whole way	107.635	106.1	a nine-foot wall
CHASE-SOUTHstation 96	★ the rising west plane, capped at the chase column’s head	75.46 at X 12 → 107.635	a SLOPE	not a single figure

⚠ What is a fact here, and what is not. The ceiling height above every one of these lines is a datum fact, derived above and asserted at generation. Whether either interior wall is BUILT to the ceiling is a design choice, and no source states it. So both are drawn as open envelopes — dashed top, no top plate — and both bars are captioned LINE LENGTH, not counted Σb. The page shows what is available; it does not build it.

2 · The wall section — at an EAVE wall

Where the ceiling is the plate line, the wall section is simple, and all three lines are derived from the datum rather than typed:

line	Z	where it comes from
top of the sole plate	1.5 in	`BOTTOM_PLATE` — a dimensional 2× sole plate
underside of the doubled top plate	63 in	`PLATE_Z` 66 − `PLATE_DEPTH` 3 (= 2 × `PLATE_PLY` 1.5)
clear framed height, EAVE WALL	61.5 in	the difference — 5'-1 1/2", in a 5'-6" wall

⚠ That figure is right for WEST and EAST, and for the wall portion of SOUTH and NORTH. It is wrong for the two interior walls — see §1.

★ Register 0.3 is the rule that then sets the sills. Kids use the treehouse unsupervised from age 4, so most glass sits high — a 42 in (3'-6") sill — with the egress-capable window excepted, which register 1.2 puts around a 34 in sill.

★ Register 0.3a: the kid-safe band governs glazing that can OPEN or BREAK. It does not govern fixed plexi. Owner, 2026-08-10: “kid safe band only applies windows that can open or break (eg glass)”. The first draft applied the band to the acrylic panel, which is applying the rule to the thing that answers it — plexi was register 0.3’s own proposed mitigation for breakable glass. So the band is drawn solid on every wall that is a candidate for glass or an operable sash, and ghosted on the fixed-plexi wall, where it is a reference line and not a constraint.

3 · The six elevations, at one common scale

The three walls at a constant station run across the ridge, so a roof triangle stands above each of them; of the three at a constant X, two run under the eaves at constant plate height — and one does not. Each panel is plotted against the plan’s own axes (X increasing east; station increasing south), not oriented to a viewer, so the south gable reads exactly as it does in the plan study, east on the right.

How to read it. The pale blue band is the candidate glazing band — the 21 in between the 42 in kid-safe sill (solid red) and the underside of the top plate. It is a zone, not an opening: no window on this page has a width, a position or a head. The dashed red line at 34 in appears only on the two walls the owner named as egress candidates. Tan bands are the plate stack; tan diagonal hatch is solid wall; the dashed tan outline on the tall interior panel marks the trunk standing behind it, 36.5 in across and centered on that wall’s own midpoint. Amber crosshatch = the gable pair at the owner’s 18 in working width; green dotted = a vent zone, zone only.
⚠ The red hatched block on the east wall carries ONE dimension only: its ~30 in width, from register 1.2a. Its height is not ruled by anything, and it is shown sliding because its position in that wall is open too.

⚠ Four things the drawing says that a table would not.
1 — the two interior walls are drawn OPEN at the top, because the ceiling above them is known and their built height is not. One of them has a top that is a slope, not a line: it rises with the west plane and caps where it dies into the chase column’s head. Neither is shown as counted racking length.
2 — the NORTH gable has only ONE side. The wall runs X 66.5 → 108, and the ridge post occupies X 66.5 → 70.5 — so the post stands at this wall’s west end, not in the middle of it. A symmetric pair is not available here; the south gable is the only one where 1.2c’s pair can be symmetric.
3 — the east end of the interior E–W wall lands on the chase column, X 66.5, which datum v7 merged with the S-mid ridge post. That is the “frame attachment point” the owner named, and it is a real member.
4 — the tall interior wall’s ceiling is the ridge beam itself, drawn across the top of that panel. Its west face and the wall line are the same plane (CHASE_E_WALL_X = RIDGE_ARRIS_W), which is exactly why that wall gets the ridge’s height and the eave walls do not.

4 · Wall × what can go in it

wall	what it does	GLASS — candidate zone	A WAY OUT	VENT
SOUTHstation 138 · 108.0 in	gable end — and the loft end	the 21 in candidate band, PLUS the gable pair above the plate	★ a named egress candidate (owner: “S wall under loft”) — siting OPEN	● gable peak — the true exterior gable
NORTHstation 43.5 · 41.5 in extent	the door wall — structurally free	the 21 in candidate band, and a gable slice with only ONE side	○ ruled OUT as a second exit — register 0.1b: it shares the door’s own failure mode (the same porch)	● gable peak — sheltered, borrowed light (register 1.2)
CHASE-SOUTHstation 96 · 54.5 in of line	interior — the loft’s north boundary; its top is a slope	⚠ nothing sized: the ceiling above it is known, its own BUILT height is not	○ interior — it opens on nothing	○ interior
EASTX 108 · 94.5 in gross	BEARING — stands over the east drop beam	the 21 in candidate band, minus wherever the exit lands	● the second exit, ~30 in (register 1.2a, DECIDED) — and a named egress-window candidate. Position OPEN	○ no triangle — it runs under the eaves
WESTX 0 · 54.0–68.25 in	BEARING — stands over the west drop beam	the 21 in candidate band — over a length that is still a RANGE	○ not named by the owner — the ledge outside is ~10 in	○ no triangle
CHASE-EASTX 66.5 · 52.5 in	interior — the tree side, and the tallest wall in the room	★ the whole wall is the candidate — the plexi wall, register 1.3a. The kid-safe band does not govern it (0.3a)	○ interior	○ no triangle — the ridge beam is its ceiling

● = the register says so · ○ = it does not, or the wall cannot · ★ = the finding worth reading. Every “candidate” on this page is a candidate; none is a selection.

5 · Chase-east — the one wall that gets a product

Register 1.3a makes the chase east wall a plexi wall — it is the tree view. Register 1.3b names the product the owner found: OPTIX 0.118 in × 36 in × 72 in clear acrylic, ~$110, Lowe’s #11289. And §1 changed what this wall is: it stands at RIDGE_ARRIS_W, directly under the ridge beam, so the opening available is 52.5 in long × 106.1 in clear.

★★ That is a 52.5 × 106 in opening onto the trunk — 38.7 ft². The tree view, rather than a ribbon. The first draft drew this wall at the eave-wall height and made the plexi a clerestory strip. At its real height it is a nine-foot wall of glazing looking straight at the oak. The owner’s own instinct — one big upright sheet — is the right shape for it, and the arithmetic below says exactly how far one sheet gets.

★ The lead case: ONE sheet, 36 in along the wall × 72 in tall

Owner, 2026-08-10: one upright sheet “gives full view of trunk”, and side-by-side sheets “will be unsightly unless we frame the joint plus poses a weather barrier issue”. So on this page a seam is a cost, not a neutral option. Two things about that single sheet are worth stating exactly:

★ “Full view of trunk” is literally true. The trunk is 36.5 in across (DERIVED — field verify) and its centerline sits at station 69.75, which is this wall’s own midpoint. A 36 in sheet centered on the wall is therefore centered on the tree, and shows 98.6% of its width.
⚠ The height is not what it looks like. 72 in of a 106.1 in wall is 67.8%, not the ~80% assumed.

⚠ The two stated preferences cannot both hold, and this page does not choose between them.
★ “~80% of the chase height” is 84.9 in of glazing.
★ One seamless 72 in sheet gives 72 in.
→ a 72 in sheet is 12.9 in short of the 80% target. One sheet or 80% — not both. Buying the last 12.9 in costs a seam and at least one more sheet; the trade is drawn below, and it is the owner’s call, not this page’s.

★ “Perfectly frame the trunk” has an exact answer — and it is no

An equal margin on all four sides is impossible with this sheet, and it fails in two different directions:

At 36 in wide the side margin is 8.25 in each side. To get the same margin top and bottom, the sheet would have to be 89.64 in tall — taller than the 72 in it is.
At 72 in tall the margin is 17.07 in top and bottom. To match that at the sides, the opening would have to be only 18.36 in wide — narrower than the 36.5 in trunk, so it would stop being a full view of the tree at all.

★★ The resolution worth drawing: equal 8.25 in on LEFT, RIGHT and TOP, with the remaining 25.89 in solid BELOW. Three sides match and the fourth is deliberately different, which reads as a framed opening with a plinth rather than as a mistake — and the solid base is the part children and furniture actually hit. Its sill lands at 27.39 in. The centered alternative (17.07 top and bottom, 8.25 at the sides) is drawn beside it so the two can be compared. Neither is selected.

option	sheets	cost	geometry	what it buys, and what it costs
P · one sheet, PLINTH	1	~$110	36 × 72 upright, 8.25 margin left, right and TOP, 25.89 solid below	★★ no seam, and it reads as a deliberate framed opening with a plinth — the solid base is the part children and furniture hit
Q · one sheet, CENTERED	1	~$110	36 × 72 upright, 17.07 top and bottom, 8.25 each side	★ no seam — but nothing lines up with anything: four margins, two values
C · two UPRIGHT	2	~$220	two 36-wide sheets side by side = 72, 19.5 trimmed; covers 72 of the height and leaves 34.14 bare above	⚠ one vertical seam — and the owner has said a seam is unsightly unless framed, and raises a barrier question
D · three LAID LONG	3	~$330	three rows 36 tall × 72 long = 108, 1.865 trimmed off the stack and 19.5 off the length of each row	⚠ two horizontal seams — it covers the whole wall, and it is the most seams of any option here

And the same wall, if it is only built to the plate line

Whether this wall is framed to the ceiling or stopped at the ordinary wall height is not stated by any source (§1). If it stops at the plate line, the sheet meets a much shorter opening, and these are the three cases — with option A now un-blocked, because the rule that used to rule it out was register 0.3’s kid-safe band, which 0.3a says does not govern fixed plexi:

A — one sheet laid long, sill at 27.0 in, 19.5 in trimmed off its length; it stands 36 in tall and 15 in of it sits below the 42 in line, which for a fixed plexi panel is now a description rather than a fault. A′ — the same sheet ripped to 21 in so it sits exactly in the band, worth doing only if the panel is ever glass or an operable sash, and it throws 15 in away. B — two stacked, 10.5 in trimmed, and one horizontal seam, which is now a cost.

⚠⚠ THE PANEL IS UNSIZED, AND THIS PAGE DOES NOT SIZE IT — and the taller it gets, the more that matters.
Whether 0.118 in acrylic may run this far without intermediate support is not established by anything on file. A 72 in tall unsupported acrylic panel is a bigger unverified question than the low ribbon the first draft drew, not a smaller one, and going to the full 106.1 in enlarges it again. Intermediate framing may well be required — that is unknown, and unknown is not a no.
Nor is it established whether a plastic panel in this location satisfies — or defeats — any safety-glazing requirement, nor how it behaves when a child runs into it. The opening is drawn; the panel is not. No figure for how far it may run, how much it deflects, or what it takes on impact appears anywhere on this page, and the generator refuses to write one.
⚠ The owner’s saved image (design-and-reference/plexi 2 sliding pane window.avif) is a marketing photo, not literature. It cannot close any of these. What can: the OPTIX / Plaskolite technical datasheet for 0.118 in clear acrylic.

⭐⭐ RULED 2026-08-10 — the trunk is NOT sealed to the roof (owner, verbatim: “we’re not sealing the trunk”; register R-1a). That is not a detail, it changes what this wall is. The chase is an open-air shaft: weather enters it by design. So CHASE-EAST is an EXTERIOR assembly, and the acrylic panel and its perimeter are the building envelope — this is a window onto open air, not an interior view onto a trunk.

⚠ Which raises the stakes on the unsized panel rather than settling them. It was an unverified structural span; it is now also the weather envelope of a wet shaft. Nothing on this page sizes it, claims it sheds water, or claims it does not.

⚠ Consequence to be designed, not a solved detail: rain and snow enter the shaft deliberately and drain down through the floor cutout, which is open around the trunk. No source states that path — it is stated here as a consequence of the ruling, and it needs designing.

⚠ Still OPEN — register R-1b: not sealing is not the same as no roof detail. The opening still needs its own geometry and an upstand or curb so water sheds INTO the shaft rather than back into the roof build-up. The owner’s working idea is to match the floor cutout — doubled trimmers and headers, on the rake. ⛔ Owner-HIL and a Fable datum pass; research-or-raise, never derived. Nothing is drawn for it here.

6 · The south gable — two equal rectangles, and where the area peaks

Register 1.2c: a symmetric pair about the S ridge post (X 66.5 → 70.5), not run to the eaves. The owner asked to see this as two equal rectangles at several widths rather than as the triangle above. Both rectangles are the same height, and that height is set by whichever slope runs out first — always the east one, because it is the steeper (53.82° against 38.25°) and its run to the wall line is only 37.5 in.

★★ The area of each rectangle PEAKS at 18.75 in wide — 25.63 in tall, 481 in² — and falls away on both sides. Past the peak, widening the pair loses more height than it gains width. So 24 in fits, and is a SMALLER window than 18 in (443 against 480 in²). The owner’s 18 in guess is within 0.2% of the best available — and 12 in gives up 13%. That is the whole answer to the question, and it is not a ruling: the widths are still a sweep.

The width sweep — each rectangle, each side of the post

width of each rectangle	height (both, equal)	area each, in²	of the peak
8 in	40.33 in	323	67.1%
12 in ★ owner’s first figure	34.86 in	418	87.0%
18 in ★ owner’s second figure — crosshatched	26.66 in	480	99.8%
18.75 in ★★ the analytic maximum	25.63 in	481	100.0%
20 in	23.92 in	478	99.6%
24 in fits, and is SMALLER than 18	18.46 in	443	92.2%
30 in	10.25 in	308	64.0%
37.5 in the east rectangle dies at the wall line	0.00 in	0	0.0%

★ And it costs zero Σb. The gable triangle sits above the top plate, on top of the full-height wall that carries the racking load. The ridge lands on its four posts, not on the gable wall. An opening up there takes nothing out of the counted length of any wall — the first and still the only genuinely free opening in this whole conversation.

⚠ These are the triangle’s heights, not window heights. Every figure in the sweep is the gable triangle’s own at that station — the space the construction allows, before any framing at all. The real clear space is lower: the rafter soffit planes pass through the birdsmouth seats, inboard of the wall lines. Nothing in this table is an opening size.
The triangle drawn is the datum’s slope-defining construction, the same one the ceiling profile in §1 uses: it meets PLATE_Z at exactly 0.0000 at both wall lines, rising APEX_RISE 54 in over runs 68.5 and 39.5 to RIDGE_X. The point where the two rafter top planes intersect is a different construction and misses the plate at both ends; a figure taken off it has already been published wrong once, so this generator asserts the closure and refuses to name the other point.

7 · Gable vents — north and south

Register 3.3a: the roof is a single-skin cathedral with no attic and no roof cavity, so there is nothing above the ceiling to ventilate. What needs venting is the room, and the driver is not summer heat but condensation on the underside of the metal in cold weather. The gable peaks are then simply the room’s high vents, working on stack effect — which is the right instrument in the right place, and §1’s profile is why: the warm air goes to the ridge.

So a candidate vent zone is drawn at each gable peak, above the window rectangles — green dotted on the SOUTH and NORTH elevations. Two things about it are worth saying out loud:

⚠ The peak is not one opening. The ridge post runs up through it to the ridge soffit, and the ridge beam sits end-on above that, so the zone is two slices on the south gable and one on the north.
South is sky; north is borrowed. The south gable at station 138 is a true exterior gable. The north wall at station 43.5 opens into the covered porch’s own roof volume, which is itself open to the north — a sheltered high vent with borrowed light, which register 1.2 read as a different and rather good property rather than a downgrade.

⚠ ZONE ONLY — no vent is sized on this page, and no sizing rule is quoted. How much ventilation this room needs is a code calculation and a building-science question, and no source for either is on file. Drawing a zone says where a vent belongs; it does not say how big it is. Register 3.3a is a research gate and it is still shut.

What would close the open items — sources, in priority order

1 · Building code / regulation. Whether a safety-glazing provision applies to a plastic panel in this location, and what it requires; whether the IRC emergency-escape provision applies to this structure at all (register 0.1 already flags this as an AHJ question, and its clear-opening figures as quoted-from-memory and unverified). Ask the AHJ; do not settle either from memory.
2 · Definitive sources — manufacturer and engineering. The OPTIX / Plaskolite technical datasheet for 0.118 in clear acrylic: its published support tables, its thermal movement figures and its fixing detail. That document, not a product page, is what makes the panel sizable — and at the heights §5 now draws, it is the single most valuable missing document on this page. The same class of source settles the single-skin condensation question in 3.3a.
3 · Trade practice. Only after the two above, and only for the fixing detail — how an acrylic panel is actually held in a timber frame so it can move.
⚠ The owner’s saved image is none of these three. It is a marketing photo of a product, and it was recorded as such the day it arrived.
⭐ One question that WAS on this list is now answered: whether this wall is inside or outside the weather line. It is outside — the trunk is not sealed (R-1a), so the chase is an open-air shaft. ⚠ What remains open is R-1b, the roof opening’s own geometry and its upstand.

Deliberately NOT drawn

Any window, at any size or position. Every glazed thing on this page is a zone or a candidate placement of a stated product. Register 1.2 is open; siting is exactly what it will decide.
Which wall gets the egress-capable window. The owner named the SOUTH and EAST walls as candidates; both are drawn with the lower sill line and neither is chosen. ⚠ That is a different question from the second exit, which register 1.2a decided (EAST) — and whose position within that wall is still open, so it is drawn sliding.
The built height of either interior wall — the ceiling above them is a datum fact and their own top is not, so both are drawn as open envelopes.
Header sizes, and any framing for any opening — jambs, sills, cripples, trimmers. None ruled, none drawn.
Stud counts, stud layout, nailing, hold-downs, sheathing type.
Any structural capacity claim for the acrylic panel — how far it may run without intermediate support, how much it deflects, how it takes an impact, and whether it is permitted where safety glazing is required. All UNSIZED; see §5.
Any weather-barrier detail or claim, for the panel, its perimeter or a seam. ⭐ This wall is now known to be outside the weather line (R-1a — the trunk is not sealed), which makes the detail more load-bearing, not less: the panel is the envelope. Nothing here details it, and R-1b — the roof opening and its upstand — is still open.
The drainage path out of the shaft. Water entering the open chase falls through the floor cutout, which is open around the trunk. That is a consequence of R-1a, stated so it is not forgotten — no source designs it, and nothing here sizes or details it.
Vent sizes, and any ventilation sizing rule. Zone only; see §7.
Shear-wall selection, and any counted racking length. Register 1.4 is reopened and handed to 4_ENG_SW. Every wall line on this page is a candidate; the 64.5 in figure quoted for the east wall is the plan study’s own candidate arithmetic, not a credit.
The west nook’s depth — register 2.2b is open, so the west wall is drawn as a range (54.0 to 68.25 in) and never as a number.
The north door’s leaf size. Register 3.2 is open and the roof does not force it; the door is drawn as a zone.
Blocking. That is register 2.1 and it has its own study — but note that every sill and head on this page will eventually want backing, and the bands were chosen before any of these openings were.

Built from geometry.py at DATUM v7 (2026-08-08c) — the station chain, the plate stack, the slope-defining triangle, Z_RIDGE_BOT and Z_RIDGE_TOP, the ridge posts and the merged chase column — plus the interiors register 2026-08-10 (rows 0.1b, 0.3, 0.3a, 0.4, 1.2, 1.2a, 1.2c, 1.3, 1.3a, 1.3b, 1.4, 1.4a, 1.5a, 2.1, 2.2a-rev, 2.2b, 3.3a, 3.4a, 3.8a, R-1) for the figures the datum does not own: the ~30 in east exit, the two sill lines and their scope, the gable-pair framing, the nook depth options and the OPTIX product. The six lengths are cross-asserted against the companion wall-by-wall plan study at generation (build_study.py::_check()), the ceiling profile is re-derived and checked against the owner’s own cross-check values, and the composed page is scanned for forbidden claims before it is written (_check_text(), _check_wall_heights()); any of them failing stops the build rather than shipping a drawn guess. • FROZEN 2026-08-10. Do not regenerate. A study records what was known when a question was framed; re-running it against a later datum produces a different document and destroys that record. • This page is never to be measured from.

📁 archived · blocking-bands — click to view

⭐ RULED R17 (2026-08-13) — the owner's HALF-DROP LATTICE supersedes this study's three-band recommendation: rows every 16 in per bay, alternate bays dropped 8 (odd bays 16/32/48, even 24/40), the rough 4×4 plate as the top band as built. The unknown-future-fixtures dilemma dissolved by pattern, not prediction.
⛔ FROZEN at its own datum · 2026-08-10 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-10-blocking-bands/

STUDY — NOT PART OF THE DESIGN — A RECOMMENDATION, NOT A RULED SCHEDULE — HEIGHTS ARE OWNER JUDGMENT, NOT A SOURCED STANDARD

Where does blocking go, when nobody knows yet what will hang on the walls?

Interiors register 2.1 · ⏸ OPEN — this page is the decision aid, and the decision is the owner’s. Wall carcass drawn from geometry.py at datum v7: plate line PLATE_Z 66 in (LOCKED), wall line WALL_SPAN 108 in, sole plate 1.5 in, doubled top plate 3.0 in → underside at 63 in, stud 1.5 × 3.5 dimensional at 16 in o.c. (register 1.1 + 1.1a-rev), clear bay 14.5 in.

⛔ The deadline arrives long before the decision feels urgent. Blocking is nearly free while a stud bay is open, and is demolition afterwards. The bays close when sheathing or interior finish goes on. So this has to be decided for things that will not be built for years — which is exactly the kind of item that gets deferred forever because nothing is pressing.

The proposal — bands, drawn against a child rather than against a wall

Three continuous heights on every wall. Two more are already there for free: the sole plate and the doubled top plate are continuous backing whether anyone plans for them or not. A fourth band is drawn as a candidate, visibly different, because that is the actual open question.

height	what it is for	why that height
0″	sole plate	already there — free
16″	bench seat, low shelf	a small child’s seat height
34″	desk, work surface, mid shelf	a four-year-old works here
52″	high shelf, hooks, picture rail	top of a young child’s reach
~58–60″	candidate fourth band	deliberately above a young child’s reach
63″	top plate underside (`PLATE_Z` 66 − 3)	already there — free

⚠ The three heights are owner judgment (register 2.1), not a sourced standard. No anthropometric table is on file in this repo, so none is quoted here and the figure above is indicative only. Stating that is what makes the heights reviewable.

1 — Design for the occupant who was never told

This is the whole argument, and it is not really about carpentry. With bands, a future person hangs something anywhere along a line without knowing where the blocking is — because it is everywhere at that height. With points, they must have read a document. The building will outlive the document, and it will outlive the conversation that produced it.

Points also demand something impossible: that the entire interior be decided now. A shelf that moves eight inches has nothing behind it.

2 — The rule that actually sets the size

THE GOVERNING RULE “Blocking is not sized for the object. It is sized for a child hanging off it.”

A shelf holds books; the same shelf holds a child who decided it was a ladder rung. A hook holds a coat; it also gets used as a foothold. Foreseeable misuse is a design input, not user error.

⛔ And that is a sizing principle awaiting a real check — not a number this study may supply. Nothing in this repo states a shelf or blocking capacity, so no member size, no fastener and no pounds figure appears anywhere on this page. The bands above are drawn as lines for exactly that reason. Sizing is a structure-decisions §5 research gate.

3 — Write the heights into the building itself

Documents get lost. The structure does not. The band heights should be recorded on the building — marked inside a hatch, written on the framing before it is closed, or carried on the names plaque (register 3.15). That costs a pencil and it is the only part of this scheme that survives everyone who was in the room.

4 — The fourth band, and the question it actually raises

The fourth band earns its place only if you want a zone a small child cannot reach — in a room children use unsupervised from age 4 (register 0.3), somewhere to put things they should not get is a real safety instrument, not just storage.

⚠ But look at what the drawing shows. The top plate underside is already continuous backing at 63 in, free. A band at 58–60 sits only 3–5 in below it. So the real question is not “is an out-of-reach zone worth having” — it is whether 63 in is already high enough to be that zone, in which case the fourth band buys a few inches of convenience rather than a new capability. That is the owner’s call, and it is why the band is drawn as a candidate and not as a proposal.

5 — Beyond the bands: four things that are not band-shaped

Listed as open. ⛔ None is drawn as a detail — no ladder design, no grab-handle detail, no hinge specification appears in this study.

The loft ledger ends. The ledger lands on the south wall and hits every stud; the ends may want blocking. ⚠ And the drawing found a collision: the loft top is LOFT_TOP_Z 54 in with a frame member LOFT_FRAME_D 3 in deep, so the ledger zone is 51–54 in — and the 52 in band sits inside it. On that wall the two want the same height. Not resolved here; the ledger detail is not drawn.
Door hinges and any closer. Solid backing. Register 0.1 makes this life-safety, not preference: the door must open from the inside without a key, tool or special knowledge. A hinge screwed to nothing is that requirement failing quietly.
⭐ A grab handle at the top of the ladder. Nobody had raised this. Permanent open ladder, no lock, unsupervised from age 4, arriving at deck level in the dark. It is the same class of item as the door, and the backing for it is free today.
Anything hung from the rafters — swing, hammock, hanging chair. ⛔ A different problem entirely, and it must not be folded into this one: that is a point load at mid-span on a dimensional 2×6 at 16 in o.c., not a wall-blocking question. It goes to the engine, not to a band.

What this study deliberately does NOT draw

omitted	why
Any blocking member size, species, fastener or nailing	⛔ No source in this repo states one. Bands are drawn as lines.
Any capacity — pounds, per-band or per-hook	⛔ Nothing in this repo states a shelf or blocking capacity. The “child hanging off it” rule is a sizing principle awaiting a real check.
Any reach or anthropometric standard	⛔ No table is on file. The heights are owner judgment; the figure is indicative and labelled so on the drawing.
The ladder, the grab handle, the door hinges/closer	Named as open items above. Detailing any of them here would be inventing.
The loft ledger itself, and its end condition	Its zone is shown because the arithmetic collides with the 52 band. The connection is unstated by any source.
Any shear or diaphragm claim for the blocking	Walls are 66 in, so a 4×8 panel runs vertically with no horizontal seam — there is no panel-edge blocking to coordinate with, and no claim is made either way.
Bands on a specific wall schedule	The proposal is every wall. Which walls are sheathed is register 1.4, reopened and handed to `4_ENG_SW`.

Two facts about the wall that the bands have to live with. The last bay on this wall is 9.0 in clear, not 14.5 — 108 does not divide by 16. And the west nook is 6 in usable of an 18 in space (register 2.2a): the trunk’s 12 in sway envelope takes the rest, and ⛔ nothing may be blocked, shelved or hung into that envelope — it is a tree strike, not a clearance.

Sources this page rests on. Wall geometry: treehouse/tools/draw/geometry.py, DATUM v7 (2026-08-08c) — PLATE_Z, WALL_SPAN, BOTTOM_PLATE, PLATE_PLY/PLATE_DEPTH, WALL_STUD_THK, LOFT_TOP_Z, LOFT_FRAME_D, RAFTER_OC. Framing module and band heights: treehouse/docs/interiors/README.md — register 1.1 (16 in o.c., DECIDED 2026-08-10), 1.1a-rev (dimensional 1.5 × 3.5, SUPERSEDING 1.1a the same day), 2.1 (the bands, ⏸ OPEN), 0.1 / 0.1b / 0.3 (the door, the second exit, age-4-alone), 2.2a (the west nook), 3.7 (sporadic shelves). Principle: the structure-decisions skill, §3 (design for the occupant who was never told) and §4 (foreseeable misuse).

📁 archived · west-6ft-wall — click to view

⭐ RULED R14 (2026-08-13) — 5'-6" FINAL; the default won. Ledger closed balanced 2-for/2-against; reopen_if the PE's enclosure/mechanism rulings move uplift or lateral materially. This stays the frozen evaluation record.
⛔ FROZEN at its own datum · 2026-08-12 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-12-west-6ft-wall/

STUDY — NOT PART OF THE DESIGN — THE 5'-6" LOCK STANDS, NOTHING HERE IS ADOPTED — GENERATED FROM geometry.py, NOT HAND-DRAWN

What would a 6'-0" wall buy the WEST ladder arrival — and what would it cost?

A what-if, asked the day the tails were tapered. The taper bought 2.86 in at the west landing and it is ruled. Wall height is the other lever, and it is inside the owner’s own ruled band — so this page draws what 6 more inches would look like, and what they would drag with them.

⛔ THE 5'-6" LOCK STANDS. This page adopts nothing. PLATE_Z = 66 is an owner lock of 2026-08-07 (2.6a §8, the kid-scale rebase) and it governs until an owner rules otherwise. Every figure here at plate 72 is a what-if: it is not a datum value, it exists only inside this page, and no sheet, workbook or memo may be measured against it.

⭐ But 6'-0" is INSIDE the band the owner already ruled. The 2026-08-07 lock’s own words, as 2.6a §8 carries them: “7' is out on the jumping concern; 5–6' is in, 5.5 as compromise”. So 72 in is not a reopening of a closed question — it is a re-choice within a band that is still open at its top end. ⚠ Mechanically it is still a real move: a PLATE_Z datum pass (Fable-only under the model policy), the workbook mirrors that quote the plate, and a full re-render of the construction set. This study does none of that and asks for none of it.

① The arrival, at 66 and at 72

How to read it. Both panels are at one vertical scale and both draw the adopted v8 tapered tail solid — the cut is ruled, so it is not drawn as an option in either. The only difference between the two panels is the plate: 2.6a §8 states that a taller wall leaves “pitch, rafter cuts, roof area, snow and the 9'-0" span all identical — the roof simply translates up”, and the generator asserts exactly that (the arrival must gain precisely 6 in, or the construction is refused).

At the west landing: 59.30 in as ruled today → 65.30 in at a 72 plate. 4'-11 5/16" → 5'-5 5/16". A 72 in adult’s duck at the deck edge goes 12.7 in → 6.7 in — a real improvement, and still a duck. The tails still overhang the landing by 5.375 in; wall height does not touch that.

② ⭐ The question underneath: why must loft headroom move with the wall?

It must not. That is the finding. LOFT_TOP_Z = 54 is an owner CHOICE, not a structural constraint — 2.12 §2b, in terms: “The loft height above the cabin floor had never been set anywhere in the project… Owner decision: 54 inches, chosen to favor the space below.” The loft is a net in a ledger frame whose edges are the three walls and the front beam (2.12 §2); the ledger ties into the S-wall stud run and can sit anywhere in it. Nothing in any source on file pins its elevation to the wall height.

So “walls +6” and “loft +6” are two independent knobs, and the jumping objection that produced the 5'-6" lock is an objection to one particular setting of them — not to wall height as such.

	TODAY — plate 66, loft 54	WALLS +6, LOFT FIXED	WALLS +6, LOFT +6
clear above the loft, ridge strip	53.635 in	⚠ 59.635 in — the jumping concern worsens, which is the exact failure mode the lock named	⭐ 53.635 in — EXACTLY today’s. Jumping unchanged.
clear above the loft, eave walls (plate line)	12 in	18 in	12 in — exactly today’s.
clear above the loft at its own E–W edges (real soffit)	12.788 W / 14.734 E	18.788 W / 20.734 E	12.788 W / 14.734 E — exactly today’s, edges included
clear UNDER the net frame	51 in	51 in — wall height cannot touch it	⭐ 57 in — “favor the space below” gets better, by the whole 6 in
climb to the loft	54 in	54 in	⚠ 60 in — 6 in more to climb, on the fixed ladder at the front beam (2.12 §2b, access RULED)

⚠ Honest secondary notes on the raised pairing.

The loft-access climb grows 6 in. Access is already ruled — a fixed ladder at the loft front beam (2.12 §2b) — so this is a longer climb for the same children, at the same place. No rung, tread or rail is specified here or anywhere.
The loft’s own edges do not pinch. Derived at its E–W extent (LOFT_STRUCT_EW 101 in, stud face to stud face): the raised pairing reproduces today’s edge clearance exactly, west and east, because the whole above-loft envelope translates rigidly with it. That is asserted, not asserted-at.
Guard and rail heights are rules, not geometry. They are measured above the walking surface, so they follow the loft, not the wall — the setting does not change them. ⚠ No rail is specified here; the guard height is still not in geometry.py (A-901’s standing escalation), and this page prints no figure for it.
⛔ It is not free. Moving the loft is a 2.12 amendment plus a LOFT_TOP_Z datum change, owner-ruled — the same class of move as the plate itself. Two datum values, not one.

③ Who stands up at the landing

What 6 more inches would cost

⛔ EVERY UPLIFT FIGURE BELOW IS QUOTED FROM 2.6a §8's OWN TABLE. No wind arithmetic is run on this page, and the rows do not share a basis — which is the point.

2.6a §8 row	net corner uplift, enclosed	partially enclosed	basis that table states
6'-0"	313 lb	444 lb	the table's own K_z 0.85 basis
5'-6" LOCKED, face derived	291 lb	426 lb	K_z 0.8688, rulings F-A + F-B
5'-6" LOCKED — superseded the same day	296 lb	430 lb	K_z 0.8688, typed 80.0 sf face

⚠ Read the bases before reading the difference. The 6'-0" row rides on the table’s K_z 0.85 — ASCE 7-16 Table 26.10-1’s 0–15 ft floor, which the 2026-08-08 rulings identified as a placeholder and corrected to 0.8688 for the locked row. 2.6a §8 says in terms that the same correction makes its 7'-0" row understated on its own basis; the 6'-0" row is in the same position. ⛔ This page does not correct it — that is new wind arithmetic, it belongs to the calcs workbook (2_LD_WIND), and a study does not author a value in someone else’s domain. The two 5'-6" rows are shown together because the table itself supersedes one with the other on the same day, and quoting either alone would misrepresent it.

⚠ The 675 lb clay comparator is 2.6a §8’s own, and so is its caveat: it is an ANALOGUE, not a rating, and a field pull test is owed regardless.

The rest of the bill, also cited.

Studs: still cut from 8 ft stock. 2.6a §8: “6'-0" and 7'-0" walls both cut from 8 ft stock” — 72 in is inside that, so the stud line does not move.
Sheathing / siding: ~38 sf, about 1¼ sheets — ⚠ that is 2.6a §8’s figure for its own 6'-0" to 7'-0" (12 in) step, not for this study’s 6 in one. Not re-scaled here: a takeoff is the BOM workbook’s and a study does not do arithmetic in another domain’s book.
Roof: unchanged. Pitch, rafter cuts, roof area, snow and the 9'-0" span are all identical — it translates.
Door and opening height allowance gains 6 in. The clear wall between the bottom plate’s top and the top plate’s underside derives to 61.5 in today (PLATE_Z − BOTTOM_PLATE 1.5 − PLATE_DEPTH 3) and would be 67.5 in — which is the height every door, window head and opening has to live inside. Register 3.2 (the door) is open.
⚠ The oak. 2.6a §8 carries an assumption on record for raising the roof: it moves the west-plane trunk penetration further up the trunk, and the owner assessed 2026-08-05 that it makes no material difference — not yet field verified. That assessment was written for a 12 in raise; this would be 6.

Deliberately NOT drawn

A recommendation, or an adoption. The lock stands; this page is an input to an owner decision and nothing else.
Any new wind, uplift or snow arithmetic. Every load figure is quoted with its basis from 2.6a §8, including the ones that document says are understated.
A corrected 6'-0" uplift row. It is derivable in the calcs workbook and it is not derivable here.
Any child stature. 2.12 §2b’s 36–48 in band is a criterion, quoted as one; the 72 in figure on the sections is register 0.4’s own adult.
Any loft rail, net edge guard or ladder detail — ruled in 2.12, specified nowhere, and not specified here.
A re-scaled sheathing quantity. Cited at the step its source states.
The ladder siting question. Still open (registers 0.4a / 0.4b / 3.2), and wall height does not settle it — see the ladder-siting study of the same date.
Anything east. The owner asked about the WEST arrival; the east eave has its own geometry and is not re-drawn here.

Built from geometry.py at DATUM v8 (2026-08-12) — PLATE_Z, both roof planes and both rafter SOFFIT planes, the v8 tapered tail (RAFTER_TAIL_TIP_D / TAIL_TIP_RISE_W), EAVE_OVERHANG_NOMINAL, the E–W deck build-up, Z_RIDGE_BOT, LOFT_TOP_Z / LOFT_FRAME_D and the three published headroom values, LOFT_STRUCT_EW, BOTTOM_PLATE and PLATE_DEPTH — plus 2.6a §8 (the wall-height decision, its uplift table and its cost list, all QUOTED) and 2.12 §2b (the loft height as an owner choice, and the kid-scale criterion band). Every drawn number is derived at generation; the generator asserts that a 6 in wall raise moves the arrival by exactly 6 in, that raising the loft with it reproduces today’s above-loft clear to 53.6350 in at the ridge and at both loft edges, and that the space below gains the whole raise — and it scans the composed page for adoption language and for invented load figures before writing it. • FROZEN 2026-08-12. Do not regenerate. • This page is never to be measured from, and the 5'-6" lock stands.

📁 archived · east-wall-openings — click to view

⭐ RULED AROUND (2026-08-13): R13 centered the exit window and R14 closed wall height at 5'-6" FINAL — dissolving the offset-door foreclosure this study flagged — and R18/R19 ruled the east program to ONE window (the E-S1b 40×24 envelope). The segment-budget arithmetic stays the record.
⛔ FROZEN at its own datum · 2026-08-13 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-13-east-wall-openings/

STUDY — NOT PART OF THE DESIGN — NO WINDOW IS SELECTED AND NO OPENING SIZE IS ADOPTED — GENERATED FROM geometry.py AND A RECALCULATED calcs.xlsx, NOT HAND-DRAWN

What can the EAST wall hold, now that it has slack?

On 2026-08-13 the owner ruled the WEST wall built — full height, sheathed, no openings. That made the north–south direction a two-line system, and the east wall’s governing segment fell from 0.903 to 0.451. It now has margin for openings it could not afford the day before. This page prices that margin.

⛔ NOTHING ON THIS PAGE IS SELECTED. Register 1.2 — the window schedule — is OPEN. The only ruled opening on this wall is 1.2a’s second exit (~30 in, EAST, 2026-08-10, PE-confirm), and even its station is not ruled: 1.2a states two layouts and picks neither. Everything else here is a dashed candidate with its arithmetic beside it.

⭐ The premise, quoted rather than claimed. 9_TH_SW_E’s own live headline, post-1f025d2: “LIVE -- EAST at the TWO-LINE split (R7): governing segment util 0.451 at v 125.6 plf, hold-down T 671 lb -- down from 0.903 / 251.1 plf / 1,362 lb when this was the only counted N-S line. WEST now carries the other half (9_TH_SW_W, util 0.357). The one remaining candidate, CHASE-EAST, would take the governing segment to 0.314 -- and it is the ruled plexi wall, so that stays priced-not-proposed.”

Basis: owner ruling R7, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R7 · CALC: 9_TH_SW_E §I2 ‘THE OPENING POSITION IS A LIVE VARIABLE’ and §I3 ‘THE N–S CANDIDATE LINES’. Both cited because a utilization printed on a page owes its rationale and its live calculation, not one of them.

① The wall as it stands

How to read it. The east wall is an eave wall, so register 1.5a rules that “the ceiling IS the plate line” — the roof rises away from it, westward. ⭐ That is the single most important fact on this page and it is a constraint: there is no gable triangle above this wall, so the free opening the S and N gables have (register 1.2, “gable-peak openings cost ZERO Σb”) does not exist here. A “clerestory” on the east wall means a high band inside the shear wall, just under the plate — and it is not free.

A high clerestory light costs exactly what a door costs: its own width. The segmented method (SDPWS 4.3.5.1, and 9_TH_SW_E’s SCOPE note says so in terms) counts only full-height segments — floor to top plate, no opening. A 24 in light 42 in up the wall breaks that segment as completely as a 24 in door does. The glass is cheap; the Σb is not.

② The candidates

③ The ledger

layout / candidate	full-height segments	Σb counted	v (plf)	governing util	what it says
A — exit CENTERED (ruled wall, unruled station)	32.25 + 32.25	64.5 in	125.6	0.380	the comfortable layout; both segments clear the aspect limit with room
B — exit OFFSET 20 in (the case the workbook binds)	20 + 44.5	64.5 in	125.6	0.451	⚠ the 20 in segment runs at 94% of the aspect limit
CL-1 — ONE CLERESTORY LIGHT, 12 in	20.25 + 32.25	52.5 in	154.3	0.551	inside the budget -- both segments survive and both still count
CL-2 — A PAIR, 12 in, one per segment	20.25 + 20.25	40.5 in	200.0	0.715	symmetric, still inside the budget -- but it is paid for twice
CL-3 — ONE CLERESTORY LIGHT, 24 in	8.25 + 32.25	32.25 in	251.1	0.761	over the budget -- the host segment stops counting ENTIRELY
W-1 — A SECOND WINDOW, 30 in	2.25 + 32.25	32.25 in	251.1	0.761	the exit opening's own width, a second time -- same Sigma-b price as CL-3

Every row re-runs 9_TH_SW_E’s own engine on its own transcribed constants (SDPWS Table 4.3A wind column 730 plf, Table 4.3A fn.3 species factor 0.91 at G 0.41, the 4.3.3 ASD divisor 2, the 4.3.4.2 Aspect Ratio Factor, the Table 4.3.4 limit 3.5). The generator refuses to write this page unless the two RULED rows reproduce the workbook’s own published utilizations to 1e−6 — they do.

⭐ The coupling that becomes decidable here

Register 1.2a’s two layouts are not equivalent under owner ruling R3. R3 (2026-08-12) left the 5'-6" plate as the lock and put 6'-0" back on the table. Aspect ratio is pure geometry, so this much can be said without touching the wind chain:

segment	h/b at the 66 in LOCK	h/b at the 72 in what-if	counts?
A — centered, 32.25 in	2.047	2.233	⭐ yes, both ways
B — offset, 20 in	3.300	3.600	⛔ no — past the 3.5:1 blocked limit

⭐ So the OFFSET layout quietly forecloses the 6-ft wall, and the CENTERED one does not. Centering the opening was already free (9_TH_SW_E §I2: “CENTERING THE OPENING IS FREE AND WORTH HAVING”); this is a second, independent reason, and it is the one that expires — once the exit is framed offset, the wall-height option is decided by accident.

⛔ What this page will NOT say: what the utilization would be at 72 in. A taller wall presents more area to the wind, so the demand moves too, and re-running that is the calcs workbook’s 2_LD_WIND chain, not a study’s. The aspect-ratio half is exact; the utilization half is not ours.

⛔ A published figure did not reproduce — RAISED, NOT FIXED

The clear framed height between plates is 60.5 in at DATUM v9, not the 61.5 in the project publishes.

PLATE_Z 66  −  PLATE_DEPTH 4  −  BOTTOM_PLATE 1.5   =   60.5 in
  (v8: PLATE_Z 66  −  2 × PLATE_PLY 1.5  −  1.5   =   61.5 in)

Datum v9 replaced the doubled 2× top plate with a single rough true 4×4 (owner: “adopt the 4×4”), and its own banner says where the inch came from: “the extra inch of plate depth comes out of the studs.” So the plate line did not move — the wall below it got an inch shorter.

Register 1.5a states 61.5 as the eave-wall clear.
Register 0.3’s sill arithmetic stacks a “top plate 3” — two plies. Its reconciled “21 in of rough opening above a 42 in sill” is now 20 in.
Register 1.3b’s “two stacked overrun the 61.5 clear by 10.5” becomes an overrun of 11.5.
The 2026-08-10 wall-elevations study guards on 61.5. ⭐ It is FROZEN at DATUM v7 and that is correct — a study is a dated snapshot and must not be re-pointed. It is not the defect; the live register rows are.

⛔ Nothing outside this folder was changed. A study reports, the owner rules, and the owning document is fixed in its own commit (the discrepancy protocol). The figure is small and it lands on every sill and head on every wall, which is why it is stated here rather than absorbed.

What else the east plane already carries

Headers are cheap on this wall, and that is measured. The ruled exit’s header runs at a governing utilization of 0.181 — NOTCHED SHEAR @ combo 3 D+S. A second opening of similar size is a header question the workbook has already shown to be comfortable. ⛔ That is NOT a header size for any candidate on this page: no source states a header depth for these openings, and register 0.3’s own reconciliation says so outright. Every candidate’s rough opening here is drawn to the top plate’s underside, with whatever header the engine requires still to come out of it.
Basis: register 1.2a (2026-08-10, PE-confirm) · CALC: 9_TH_HDR_EXIT_E, GOVERNING UTILIZATION / §HDE1.
Outside this wall is a ledge, not a landing. Register 0.1b rules egress here means a second way out of the ROOM, onto a ledge you stand on and wait — “a refuge, not a route”. It gives the east ledge as 9.6 in.
⚠ Raised by the 2026-08-12 ladder-siting study and still open: derived from the E–W build-up the ledges look swapped — JOIST_OVERHANG_EW 5.625 + SIDE_RIM_W 4 = 9.625 outside each beam centerline, and the 0.75 in of wall-versus-beam slop falls at the EAST, making that one 10.375. This page does not re-decide it; either reading leaves a ledge, and neither is a place to stand comfortably. ⚠ FIELD VERIFY — one tape across the deck at this wall settles which reading is right, and no ledge figure is drawn as a dimension anywhere on this page. No *_NOMINAL datum value is printed here at all, so nothing else on the page carries that flag.
The kid-safe sill band still applies here. Register 0.3a: the band governs glazing that can open or break. Every candidate on this page is glass or an operable sash, so all of them sit at or above the 42 in line except W-1, which is drawn lower as a candidate for an egress-capable opening — 0.3’s own stated exception. ⚠ That exception is 0.3’s, not this page’s, and the S wall is where the ruled egress window lives (1.2f) — see the companion study.

⚠ Where this arithmetic could be wrong in the wall’s favor

The engine is SEGMENTED-ONLY, and it says so. 9_TH_SW_E’s SCOPE note: “SEGMENTED WALLS ONLY (SDPWS 4.3.5.1…). No perforated shear wall (4.3.5.3…) and no force-transfer-around-openings wall (4.3.5.2).” Both of those methods exist precisely to credit the sheathing above and below an opening, which is most of what a high clerestory leaves behind. ⭐ So the prices on this page are the conservative ones, and a clerestory could be cheaper than it looks here — but not by an amount anyone may assume. Choosing a method is an engineering decision with its own conditions and its own detailing; it is a PE and workbook question, and a study does not make it. It is named here so the conservatism is visible rather than silent.

⚠ And one that could run the other way: this wall’s share of the N–S shear is held FIXED. Under ruling R7 the two N–S lines each carry half, and the workbook’s SCOPE note says in terms: “NO SHEAR-WALL-LINE DISTRIBUTION (4.3.3.4) — V arrives at C4 already assigned to this line.” So every candidate on this page changes this wall’s Σb and nothing else. A real distribution would move force between the lines as one of them is opened — which direction it moves depends on the diaphragm and is not ours to assume. Named so the assumption is visible.

Deliberately NOT drawn

A choice, a recommendation, or an adopted size. Register 1.2 is open and this page is an input to it.
The exit opening’s station. 1.2a rules the wall and the approximate width; the two layouts are its own and neither is picked.
Any header depth — unsourced for these openings (register 0.3’s own reconciliation), so every rough opening runs to the plate underside and the header comes out of it later.
Any glazing product, sash type, operation, hardware or glass specification. Register 1.3 (the plexi barrier) is parked, 1.3b’s product gate is open, and none of it is decided here.
Any code-compliance verdict. The SDPWS numbers are the workbook’s transcription and the utilizations are ours, not a licensed engineer’s.
New wind arithmetic of any kind — including the utilization at a 72 in plate.
A perforated or force-transfer-around-openings treatment. Named above, not run.
Hold-down hardware. The tensions the workbook reports are demands; no device is selected anywhere in this project, which is why 9_TH_SW_E’s deflection is flagged NOT QUOTABLE and is not quoted here.
Venting. Register 3.3a is a shut research gate; a clerestory that opens is a vent, and that is exactly why it is not sized here.

References

(1) Code and standards — ANSI/AWC SDPWS-2015 ch.4: Table 4.3A (nominal unit shear, wind column; footnote 3 specific gravity adjustment), 4.3.3 (ASD reduction factor 2.0, “no further increases”), Table 4.3.4 (maximum shear wall aspect ratios; blocked wood structural panels 3.5:1), 4.3.4.2 (Aspect Ratio Factor (WSP) = 1.25 − 0.125 h/b_s), 4.3.5.1 (segmented method). ⛔ Every one is quoted here only as transcribed on 9_TH_SW_E’s CODE BASIS block, which carries the printed page and the date it was read.
(2) Project sources of record — geometry.py DATUM v9; calcs.xlsx tabs 9_TH_SW_E, 9_TH_SW_W, 9_TH_SW_S, 9_TH_HDR_EXIT_E (read by defined name off a LibreOffice-recalculated copy); docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R5/§R7; docs/decisions/2026-08-12-siting-wall-plexi-rulings.md §R3; interiors register rows 0.1b / 0.1c / 0.3 / 0.3a / 1.2 / 1.2a / 1.4 / 1.5a.
(3) Companion studies — 2026-08-10-wall-elevations (frozen v7), 2026-08-10-wall-by-wall, 2026-08-12-west-6ft-wall (the R3 evaluation basis), 2026-08-13-south-wall-openings (the ruled egress window).

Built from geometry.py at DATUM v9 (2026-08-13) — WALL_SPAN, TRIMMER_N_STATION, RIM_NS/S_CROSS_TO_S_RIM, PLATE_Z, PLATE_DEPTH, PLATE_PLY, BOTTOM_PLATE, JOIST_OVERHANG_EW/SIDE_RIM_W/BEAM_CC_EW/ DECK_OVER_RIMS_EW, CHASE_E_WALL_X, CUTOUT_HEADER_W_X — and from calcs.xlsx post-1f025d2, read by defined name off a headless LibreOffice recalculation and never retyped. The generator asserts that its re-run of the shear-wall engine reproduces 9_TH_SW_E’s own published segments to 1e−6, that every candidate reduces Σb, and that no drawn label prints the superseded 61.5 in clear height — and it scans the composed page for selection language, invented header sizes, code verdicts and new wind figures before writing it. • FROZEN 2026-08-13. Do not regenerate. • This page is never to be measured from, and nothing on it is adopted.

📁 archived · south-wall-openings — click to view

⭐ RULED R18 (2026-08-13, the same day) — SE-3: 27×32 clear, HARD EAST, sill ≤ 44 is the second means of egress of record (the hard-east siting DELETED the ridge-post/king-stud PE question this study raised), and above the loft G-A, a single 33×26 FIXED pane west of the ridge post, took the gable-pair's place. The shear arithmetic (height free, width spent) stays the record.
⛔ FROZEN at its own datum · 2026-08-13 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-13-south-wall-openings/

STUDY — NOT PART OF THE DESIGN — THE EGRESS WINDOW IS A RULED REQUIREMENT AND ITS SIZE IS NOT — NO OPENING SIZE IS ADOPTED — GENERATED FROM geometry.py AND A RECALCULATED calcs.xlsx

The SOUTH wall — above and below the loft, and the egress window

One wall, cut in two by the loft, with opposite economics on either side of the top plate. Above it the gable triangle is not part of the shear wall at all, so an opening up there is free. Below it is the shear wall itself — and that is where the owner has ruled a window.

⭐ RULED 2026-08-13 (owner), register 1.2f: ONE window in the SOUTH wall OPPOSITE THE NORTH DOOR (door X 66.5–108), sized for emergency exit, in the below-loft zone, exiting onto the 6 in south ledge.

⛔ What is ruled is the REQUIREMENT. The SIZE and the POSITION are OPEN, and this page adopts neither. Register 0.1b rules that egress sizing here is “a research gate, not the code figure by default” — so the reference figures below are marked [REG] REFERENCE, NOT ADOPTED and nothing on this page selects a window.

⭐ Why the window is affordable at all, quoted rather than claimed. 9_TH_SW_S’s own live headline, post-1f025d2: “LIVE -- TWO-LINE E-W STATE (owner ruling R10): SOUTH carries HALF the E-W base shear at v 97.4 plf of 332.2 plf allowable, util 0.293, h/b 0.61, hold-down T 430 lb -- down from util 0.587 / T 966 lb when it was the only counted E-W line. CHASE-SOUTH carries the other half: util 0.581 at v 193.1 plf, T 1,671 lb (9_TH_SW_CS). *** WHAT THE SECOND LINE BOUGHT IS UNIT-SHEAR RELIEF AND REDUNDANCY, NOT TORSION: the two E-W lines sit on the SAME side of the load line, so E-W torsion falls only 23.1% (2_LD_WIND 14.4), against 94 % in N-S where EAST and WEST straddle it.”

Basis: owner ruling R10, memo docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R10 · CALC: 9_TH_SW_S §I2 (the CHASE-SOUTH candidate row, now superseded by the ruling) and its RESULT block. Both cited because a printed utilization owes its rationale and its live calculation.

① The wall, floor to apex

Above the top plate an opening is free. Below it, it costs its own width. Shear walls are the full-height segments floor → top plate; the rake wall above is a separate element. So the gable triangle (54 in of it) costs zero Σb — and the 12 in strip between the loft top and the plate does not: it is still shear wall, and an opening there is priced exactly like one below the loft.

② The free half — the gable pair

This half is effectively settled, and it was settled by arithmetic. Register 1.2c rules the shape (two equal rectangles flanking the S ridge post, X 66.5–70.5 — not run to the eaves, not the triangle above) and 1.2d ruled the arithmetic: the area of each rectangle peaks at w = 18.75 in, h 25.63, 481 in² each, because

area = w ×
tan θ_E × (37.5 − w)

and that maximizes at half the east run. ⭐ 24 in fits and is a SMALLER window than 18 in.

⚠️ Two true statements that look like a contradiction, reconciled: register 1.2d says the owner’s 18 in guess is “within 4%” of optimal and the 2026-08-10 wall-elevations study says “within 0.2%”. Both are right and they measure different things — 18 is 4% off the peak by WIDTH and 0.2% off it by AREA. The curve is flat at its top, which is the useful fact: the width choice barely matters, so choose it for looks, framing or venting.

⛔ Not settled and not drawn: whether there is glass, a sash, a vent or nothing; any vent area or sizing rule (register 3.3a is a shut research gate); the head detail where the rectangle meets the rake. The triangle is register G-1’s slope-defining construction — apex on RIDGE_X 68.5, rise 54 — which the generator asserts closes on PLATE_Z at exactly 0.0000 at both wall lines. ⛔ Never APEX_X, where the rafter top planes meet: G-1 records that that construction has already produced one wrong published figure.

③ The costly half — the ruled egress window

[REG] REFERENCE VALUES — NOT ADOPTED. IRC R310 ‘Emergency Escape and Rescue Openings’ (R310.1.1–R310.1.4 in the 2015 / 2018 / 2021 editions):

quantity	reference figure	what it does here
net clear opening AREA	5.7 sf (820.8 in²)	at the reference minimum height it needs 34.2 in of width — nearly the whole 41.5 in ruled stretch
net clear opening HEIGHT, min	24 in	fits: the below-loft clear is 51 in
net clear opening WIDTH, min	20 in	⭐ at this width the area figure needs 41.04 in of height — which fits under the 51 in below-loft clear, with the sill down at 9.96 in. That is the cheapest route on `Σb` and the most awkward one to live with (SG-2)
sill height above floor, MAX	44 in	every candidate is far below it (sills 9.96–27)

⛔⛔ These are NOT this project’s target and they are not adopted here. Register 0.1b RULED (2026-08-10) that “egress is just any 2nd exit option other than front door” — a redundancy requirement, not a code-compliance one — and kept the IRC arithmetic “on file as the UPPER bound; it is no longer the target”, with sizing named a research gate. Register 0.1’s own open items add two more reasons not to lean on them: the figures were “quoted from memory to test feasibility — verify before any of them is relied on”, and whether the provision applies at all is an AHJ question (it governs sleeping rooms in dwelling units; an accessory structure likely is not one). ⚠️ The section number is edition-dependent and the local AHJ edition is unconfirmed — the same open item the J10 ridge-strap note carries.

⭐ The one thing the reference set does say clearly, and it is useful: its four figures do not combine at their minima. 20 × 24 is 3.33 sf, well under 5.7 sf. Width and height minima are floors, not a specification, and the area is what actually sizes the hole.

On a shear line, HEIGHT is free. Only WIDTH is spent. SG-1 (20 × 24) and SG-2 (20 × 41.04) cost the shear wall exactly the same — Σb 88 in, util 0.416, identically — because the segmented method counts the length of full-height segments and nothing else. SG-2 buys 1.7× the opening area for free. ⚠️ Its price is paid somewhere else entirely: a sill at 9.96 in is essentially at floor level, which is a different conversation about a room used by a four-year-old alone (register 0.3) — and one this page raises rather than settles.

④ The ledger

state / candidate	rough opening	counting segments	Σb	v (plf)	governing util	what it says
as it stands — no opening	—	108	108 in	97.4	0.293	halved by R10; h/b 0.61, hold-down T 430 lb
SG-1	20 × 24 (3.33 sf)	66.5 + 21.5	88 in	119.6	0.416	both REFERENCE MINIMUM DIMENSIONS, drawn to price them — but their own area figure is nowhere near met at this size
SG-2	20 × 41.04 (5.70 sf)	66.5 + 21.5	88 in	119.6	0.416	⭐ the REFERENCE AREA at the reference minimum WIDTH — and it costs the shear line EXACTLY what SG-1 costs, because only width is spent. Its sill is the price instead
SG-3	24 × 30 (5.00 sf)	66.5	66.5 in	158.2	0.476	a practical middle shape — the east stub stops counting, and 4 more inches of width is what buys that
SG-4	34.2 × 24 (5.70 sf)	66.5	66.5 in	158.2	0.476	the REFERENCE AREA at the reference minimum HEIGHT — register 0.1's own '~34 in of Sigma-b' figure, and it very nearly fills the ruled stretch

Every row re-runs 9_TH_SW_S’s own engine on its own transcribed constants (SDPWS Table 4.3A wind column 730 plf, Table 4.3A fn.3 species factor 0.91 at G 0.41, the 4.3.3 ASD divisor 2, the 4.3.4.2 Aspect Ratio Factor, the Table 4.3.4 limit 3.5). The generator refuses to write this page unless the un-opened wall reproduces the workbook’s own published v and utilization to 1e−6 — it does.

⚠ Two limits on the arithmetic above, both the workbook’s own

SOUTH’s share of the E–W shear is held FIXED. Under ruling R10 the two E–W lines each carry half, and the workbook’s SCOPE note says in terms: “NO SHEAR-WALL-LINE DISTRIBUTION (4.3.3.4) — V arrives at C4 already assigned to this line.” So every candidate here changes SOUTH’s Σb and nothing else. ⚠️ A real distribution would move force between lines as one of them is opened — that is an engineering decision with its own conditions, it belongs to 2_LD_WIND and the PE, and a study does not make it. It is named so the assumption is visible rather than silent.
The engine is SEGMENTED-ONLY — no perforated shear wall (SDPWS 4.3.5.3), no force-transfer-around-openings (4.3.5.2). Both exist to credit the sheathing above and below an opening. ⭐ So these prices are the conservative ones, and the window could cost less than it looks here — but not by an amount anyone may assume.
⚠️ SOUTH is not the busy E–W line. CHASE-SOUTH runs at 0.581 — the highest of the four walls — at h 8.97 ft, hold-down T 1,671 lb at each end, “BOTH unruled applications on a trimmer substrate”. Nothing on this page touches it, and a window in SOUTH does not help it. CALC: 9_TH_SW_CS, RESULT block.

What the window runs into, stated so none of it happens quietly

⭐⭐ The ruled stretch’s west edge IS a loaded post. The S gable ridge post stands in this wall at X 66.5–70.5 — exactly where the door-opposite stretch begins — running floor to the ridge-beam soffit and carrying 271.6 lb down through a nailer and shim onto the Year-1 flush 4×10 cross. A window jamb at X 66.5 lands on it.
⛔ Whether a loaded ridge post may also serve as a window king stud is a PE question and it is NOT answered here — it changes what braces that post and adds a header reaction beside it. Flagged, not designed. Basis: 2.6a §0 / LIVING §12 (the LOCKED four-post scheme; the gable bearing chain is post → doubled 2×8 nailer → SHIM → cross) · CALC: 9_TH_RIDGE_3SPAN §5 ‘SUPPORT REACTIONS’. The seat below the floor is already scheduled J12, PE TO SPECIFY.
⭐ The loft ledger hits every stud on this wall, and an opening interrupts that. Register 2.1: the ledger “hits every stud; ends may want blocking”. Over the opening’s width the intermediate studs are cut, so the ledger has only the jamb studs behind it there. ⛔ No source states how the ledger is carried across an opening — flagged for the ledger detail, not invented. The loft frame’s own depth N–S is LOFT_DEPTH_NS_NOMINAL 42 — NOMINAL, FIELD VERIFY.
Register 3.4a lands exactly on this wall, and it helps. Interior ½ in sheathing on the S wall lands dead flush with the true-4 ridge post (post 4, dimensional stud face 3.5, so the post stands 0.5 proud — exactly the facing thickness). ⭐ Its real benefit is flatness: without it the ledger has to jog around the proud post at mid-wall. 3.4a is ⏸ OPEN, and ⚠️ sheathing both faces of a shear line is not simply double — a structure-decisions §5 research trigger, not a free gain.
The kid-safe sill band is deliberately broken here, and that is 0.3’s own exception. Register 0.3 elevates most glass to ~42 in “with exception of possibly the egress window”, and 0.1b says a high sill makes climbing out and down onto a narrow ledge harder for a child, not safer. Every candidate sits below the band, deliberately.
Where you land is a ledge 6 in wide, ~10 ft up, behind the Year-1 guard — and that re-derives exactly from the datum (RIM_NS 144 − station 138). 0.1b calls this class of exit “a refuge, not a route”, and that is the right name for it.
⚠️ A ROUGH opening is not a NET CLEAR opening. Frame and sash come out of it, and no source on file states that allowance. Every candidate here is drawn as a rough opening running up to the loft frame’s underside, with whatever header is required still to come out of it — and no source states a header depth for these openings either (register 0.3’s own reconciliation says so). ⛔ So no candidate can be said to meet or miss any reference figure. The east wall’s exit header runs at 0.181 and the north door’s lintel at 0.014, which is evidence that headers are cheap in this building — not a size for anything on this page. CALC: 9_TH_HDR_EXIT_E and 9_TH_HDR_DOOR, GOVERNING UTILIZATION.

⛔ A published figure did not reproduce — RAISED, NOT FIXED

The clear framed height between plates is 60.5 in at DATUM v9, not the 61.5 in the project publishes (PLATE_Z 66 − PLATE_DEPTH 4 − BOTTOM_PLATE 1.5). Datum v9 replaced the doubled 2× top plate with a single rough true 4×4 and its banner says where the inch went: “the extra inch of plate depth comes out of the studs”. Register 1.5a, register 0.3’s sill arithmetic and register 1.3b all still carry the v8 figure. ⛔ Nothing outside this folder was changed — a study reports and the owning document is fixed in its own commit. The companion east-wall study of the same date states it in full; it is repeated here because it lands on every sill and head on this wall too.

⭐ It does not move anything on this page’s below-loft arithmetic: that zone is bounded by the loft frame (51 in), not by the top plate.

Deliberately NOT drawn

A size for the egress window, or a position within the ruled stretch. 1.2f rules the requirement; the study prices candidates and selects none.
Any adoption of a code figure. 0.1b rules sizing a research gate; IRC R310 appears only as [REG] REFERENCE, NOT ADOPTED, with its edition and applicability caveats.
Any code-compliance verdict, for the window or the wall.
Any frame, sash, hardware, operation, opening direction, glazing product or glass specification — and no frame/sash allowance, because none is on file.
Any header depth, here or on the gable pair.
A gable-pair width. 1.2c is open; the curve is drawn so the choice can be made on looks, framing or venting rather than on area.
Any vent area, vent sizing rule or venting scheme — register 3.3a is a shut research gate and 3.3b folds into it.
The loft ledger’s detail across an opening, or any blocking layout — register 2.1 is open and the recommendation on its table is the owner’s to rule.
Any king-stud, jamb or post detail at X 66.5 — that is the PE question this page raises.
New wind arithmetic, including any redistribution between the two E–W lines, and any utilization at a 72 in plate.
A perforated or force-transfer-around-openings treatment.
Hold-down hardware. The tensions are demands; no device is selected anywhere in this project, which is why 9_TH_SW_S’s deflection is flagged NOT QUOTABLE and is not quoted here.
Anything on the EAST wall — the second exit (1.2a) has its own study of the same date.

References

(1) Code and standards — ANSI/AWC SDPWS-2015 ch.4: Table 4.3A (nominal unit shear, wind column; footnote 3 specific gravity adjustment), 4.3.3 (ASD reduction factor 2.0), Table 4.3.4 (maximum aspect ratios; blocked wood structural panels 3.5:1), 4.3.4.2 (Aspect Ratio Factor (WSP)), 4.3.5.1 (segmented method), 4.3.3.4 (shear-wall-line distribution — named, NOT applied). ⛔ Each is quoted only as transcribed on 9_TH_SW_S ’s CODE BASIS block, which carries the printed page and the date it was read. — IRC R310 ‘Emergency Escape and Rescue Openings’ (R310.1.1–R310.1.4 in the 2015 / 2018 / 2021 editions): [REG] REFERENCE, NOT ADOPTED, edition unconfirmed, applicability an AHJ question (register 0.1).
(2) Project sources of record — geometry.py DATUM v9; calcs.xlsx tabs 9_TH_SW_S, 9_TH_SW_CS, 9_TH_SW_E, 9_TH_HDR_DOOR, 9_TH_HDR_EXIT_E (read by defined name off a LibreOffice-recalculated copy) and 9_TH_RIDGE_3SPAN §5 via the datum; docs/decisions/2026-08-13-4x4-plate-and-nook-final.md §R5/§R10; 2026-08-12-siting-wall-plexi-rulings.md §R3; build-guide/phase2/2.6a §0 and 2.12 §2b/§9.2; interiors register rows 1.2f / 0.1 / 0.1b / 0.3 / 0.3a / 1.2 / 1.2c / 1.2d / G-1 / 1.4a / 1.5a / 2.1 / 3.3a / 3.4a.
(3) Companion studies — 2026-08-13-east-wall-openings (the same day, the other wall), 2026-08-10-wall-elevations (frozen v7 — the gable-pair arithmetic reproduced here), 2026-08-10-wall-by-wall, 2026-08-12-west-6ft-wall.

Built from geometry.py at DATUM v9 (2026-08-13) and from calcs.xlsx post-1f025d2, read by defined name off a headless LibreOffice recalculation and never retyped. The generator asserts that the gable triangle closes on PLATE_Z at 0.0000 at both wall lines, that the east plane governs the pair’s height at every swept width, that the area peaks at half the east run, that the un-opened wall reproduces 9_TH_SW_S’s own v and utilization to 1e−6, that every candidate lies inside the RULED stretch and below the kid band, and that the reference figures do not combine at their minima — and it scans the composed page for selection language, adopted code figures, invented header or frame allowances, compliance verdicts and new wind figures before writing it. • FROZEN 2026-08-13. Do not regenerate. • This page is never to be measured from; the requirement is ruled and no size is adopted.

📁 archived · porch-trunk-view — click to view

⭐ RULED R36 (2026-08-15) — Option B confirmed: the door-wall extension to X 42 (sta 43.5, X 42→66.5), full height, sheathed, adopted pending PE; new solid 4×4 end post at X 42; nothing at Z 66 west of X 42, ever. Torsion −22.5 % against 56 % of the trunk hidden from mid-porch — the trade the owner came here to SEE.
⛔ FROZEN at its own datum · 2026-08-13 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-13-porch-trunk-view/

STUDY — NOT PART OF THE DESIGN • ⏸ OPEN: NOTHING IS SELECTED • FROZEN AT DATUM v9 (2026-08-13) • NEVER TO BE MEASURED FROM

The door-wall extension: what it buys, and how much trunk it hides

The cabin's north wall line is the last wall lever the E–W direction has. Extending it west adds shear length on the far side of the wind resultant — and puts sheathing between you and the oak. This page prices both sides of that trade and chooses neither.

THE OWNER'S ASK, VERBATIM — 2026-08-13 “need to see studies of these options to visualize before decision is made; want tree trunk exposed experience while on porch but a portion can be concealed by wall extension if it buys structural integrity.”

The wall tops out at the plate (66 in). The oak carries on to the roof soffit — 92.6 in above the deck at its center-line. Everything you keep is in that gap. And the gap is smaller than it sounds, because the oak stands only 8 in behind the wall: from mid-porch at a 60 in eye, the maximum extension leaves 27% of the visible trunk face showing. The shorter one leaves 44%.

⏸ Ruling state — nothing here is selected

The owner deferred this decision in order to see it drawn. Three extension lengths are on the page as candidates: none, a 24.5 in segment to X 42, and a 42.5 in segment to X 24. None is recommended here, and this page recommends nothing. The register rows it informs — 1.4 (which walls are sheathed) and 1.2 (the window schedule, since a port in this segment is a variant) — stay open. A ruling on it is an owner call, and the arithmetic then belongs to calcs.xlsx, not here.

The setup, in three facts

The wall line. Station 43.5 — the N double trimmer. The cabin's north wall runs X 66.5–108 today and carries the north-facing door. West of X 66.5 there is no wall and no header: the roof spans on over the covered porch, and that open slot is the tree view.
The load line. Station 77.0 — the centroid of the E–W loaded face, which runs station 16 to 138 because the roof carries north over the porch even though no cabin wall does (calcs.xlsx · 2_LD_WIND section 11e, name wind__load_cx_ew). Both counted E–W walls — SOUTH at 138 and CHASE-SOUTH at 96 — sit south of it, by 61 in and 19 in. Nothing counted sits north of it.
The oak. X 18–54.5, stations 51.5–88, ∅36.5 in derived — FIELD VERIFY at deck height. Its north face stands 8 inches south of the wall line, not eight feet.

⚠️ FALSIFIED WHILE DRAWING — 1 of 3. The task brief described the oak as “8+ ft south of the wall line” and asked for the cutoff to be derived at “8.35 ft behind”. The datum says TRUNK_N_EDGE_STATION 51.5 against TRIMMER_N_STATION 43.5 — 8 inches, a factor of twelve. The generator refuses to build if that ever exceeds a foot, because the whole sight geometry turns on it: at eight feet a low eye would see over the wall easily; at eight inches it barely sees past it at all. ⛔ Nothing has been corrected in any source from here.

Figure 1 — the plan: what the extension covers

Three standing positions, each with the two rays that graze the oak. Where those rays cross the wall line is the oak's shadow on it — and a wall matters only where it overlaps that shadow.

⭐ The finding the plan produces on its own. From P2, by the door, the oak's shadow runs X 52.8–77.4 — entirely east of X 42. So for that standpoint, option C costs exactly what option B costs (both 72% at the reference eye) while buying 16.9% more torsion reduction. The extra 18 in of wall is free from the door. It is not free from mid-porch, and it is not free from the north edge. That is asserted in the generator, not read off the picture.

The passage is load-bearing on the ruling, and it is the softest number here. Memo R6 (docs/decisions/2026-08-13-4x4-plate-and-nook-final.md) sizes a 24 in NOMINAL ladder allowance — but it states that 24 as a N–S landing depth (band = 52.5 − nook depth, stations 43.5 → 69.75 = 26.25 in at the ruled depth). Reading it as an E–W clear width at the trimmer line, which is what bounds option C, is a transfer to a different axis. ⛔ No source on file states a passage width. It is drawn NOMINAL and FIELD VERIFY, and if it grows, option C shortens inch for inch. (Falsification 2 of 3.)

Figure 2 — the money view: looking south from the porch

Why the oak is drawn where it is

The wall lives in the plane of station 43.5. So the honest way to draw what it hides is to project the oak onto that plane from the eye, and then ask a flat question: which of the projected oak falls inside the rectangle the wall occupies? For a point on the trunk at station s, with the eye at station s_v and height h:

t = (43.5 − s_v) / (s − s_v) X_proj = x_v + t(x − x_v) Z_proj = h + t(z − h)

Two consequences fall straight out, and both are asserted in the generator. ⭐ Along any one sightline the apparent height of the oak does not depend on the eye at all — only its position does; it is exactly t × the oak's own visible height. On the ray to the trunk's north face from mid-porch, t = 0.6322, so 92.6 in of real trunk appears as 58.5 in — whether a child or an adult is looking. ⭐ And an eye exactly at the plate cuts the oak at the plate, from any distance — below it the sightline rises and you lose more; above it the line falls and you lose less. The plate height is the hinge of the whole question.

Figure 3 — the similar triangle, drawn

The closed form the section draws: Z_cut = eye + (PLATE − eye) × (s − s_v) / (43.5 − s_v). At mid-porch (station 29.75) the distance ratio to the oak's north face is 1.5818, so a 60 in eye is cut at 69.49 in = 5'-9 1/2" and a 72 in one at 62.51 in = 5'-2 1/2". Step back to the porch's north edge (station 16) and the ratio falls to 1.2909: the same 60 in eye is cut at 67.75 in instead. Backing away recovers trunk; walking up to the wall loses it.

How much oak you lose, by standpoint and eye height

Percentage of the visible trunk face — the area you can actually see, floor to roof soffit — that falls behind sheathing. The “none” column is not always zero: from the door, the existing wall already takes some.

standpoint	eye (in)	A · none	B · to X 42	C · to X 24	what the eye height is
P1 · MID-PORCH X 54, station 29.75 oak spans X 38.4–54.2 on the wall line	40	0%	68%	89%
	48	0%	63%	83%
	52	0%	61%	80%	register 2.1's top of a young child's reach — a REACH, not an eye
	60	0%	56%	73%	the page's reference eye — ILLUSTRATIVE
	66	0%	52%	69%	⭐ the hinge: eye level with the plate
	72	0%	49%	64%	register 0.4's 6 ft adult — a STATURE, drawn as an upper bound on where an eye could be
P2 · BY THE DOOR X 87.25, station 29.75 oak spans X 52.8–77.4 on the wall line	40	36%	89%	89%
	48	33%	83%	83%
	52	31%	79%	79%	register 2.1's top of a young child's reach — a REACH, not an eye
	60	27%	72%	72%	the page's reference eye — ILLUSTRATIVE
	66	24%	66%	66%	⭐ the hinge: eye level with the plate
	72	21%	61%	61%	register 0.4's 6 ft adult — a STATURE, drawn as an upper bound on where an eye could be
P3 · AT THE PORCH'S NORTH EDGE X 54, station 16 oak spans X 33.2–54.2 on the wall line	40	0%	46%	80%
	48	0%	44%	76%
	52	0%	43%	75%	register 2.1's top of a young child's reach — a REACH, not an eye
	60	0%	41%	72%	the page's reference eye — ILLUSTRATIVE
	66	0%	39%	69%	⭐ the hinge: eye level with the plate
	72	0%	38%	67%	register 0.4's 6 ft adult — a STATURE, drawn as an upper bound on where an eye could be

⛔ No child stature and no eye height is invented here, and none exists on file. Every eye height in that table is an ILLUSTRATIVE parameter, swept so the answer can be read at whatever stature matters. The only two human dimensions the project states are register 0.4's 6 ft adult — a stature, and an eye sits below it — and register 2.1's 52 in top of a young child's reach, which is a reach and not an eye at all. Neither is an eye height and neither is used as one.

The structural prize

What each candidate length would be worth in plan symmetry — and what the lever's ceiling is.

Derived with the same length-weighted center-of-resistance arithmetic calcs.xlsx · 2_LD_WIND section 14.4 uses, on the same inputs, read by name off a recalculated copy: wind__V_asd_ew = 1,753.75 lb (0.6W, ASD), wind__load_cx_ew = 77.0 in, SOUTH 9 ft at station 138 and CHASE-SOUTH 4.5417 ft at station 96. The generator refuses to write this page unless its as-built row reproduces wind__cr_ew_counted (123.9138) and wind__torsion_ew (6,856.27) to six decimals — which it does.

option	added b (in)	Σb (ft)	center of resistance (in)	residual e (in)	torsion V×e (lb-ft)	vs today	V/Σb (plf)	segment h/b	ARF
A · NONE — as today	0	13.5417	123.91	46.91	6,856	—	129.5	—	—
B · WEST to X 42	24.5	15.5833	113.38	36.38	5,317	−22.5%	112.5	2.694	0.913
C · WEST to X 24 — the maximum	42.5	17.0833	107.24	30.24	4,420	−35.5%	102.7	1.553	1.000
BOUND — the whole 108 in at this station ⛔ NOT ON THE TABLE: it walls the porch off entirely	108	22.5417	91.81	14.81	2,164	−68.4%	77.8	0.611	1.000

Option B removes 22.5% of the plan's E–W torsion. Option C removes 35.5%. For comparison: when the owner made the WEST wall countable (R7), the N–S direction's torsion fell by 94.1% — because EAST and WEST sit on opposite sides of their load line and straddle it. Nothing at station 43.5 can do that here.

⭐⭐ The ceiling on this lever, derived rather than asserted. To put the center of resistance on the load line, a wall at station 43.5 would have to be 227.6 in long — 2.1× the cabin's entire 108 in width. It is arithmetically impossible, because station 43.5 is only 33.5 in north of the resultant while SOUTH is 61 in south of it with 9 ft of length pulling. Even walling the porch off completely — the bound row, and it is not on the table — leaves 2,164 lb-ft. This lever narrows the E–W eccentricity; it cannot close it. (Falsification 3 of 3: the brief's controller envelope put option C at ≈4,429 lb-ft. This page derives 4,420, a 0.2% difference, and option B at 5,317 against ≈5,320. The derivation here reproduces the workbook's own as-built figure exactly, so it is the one carried.)

⚠️ Read what “torsion” means in this workbook before spending anything on it. 2_LD_WIND section 11 declares a flexible-diaphragm tributary distribution and states the consequence in terms: a flexible diaphragm does not distribute torsion at all, so the eccentricity computed in section 11e is not resisted by anything on that tab — it is REPORTED. So these figures are a planning-level measure of plan asymmetry, not a checked demand on anything. A rigid analysis is unavailable, because it needs a diaphragm stiffness that is not established. Reducing the number is still worth something — a better-centered plan is a better-centered plan — but nothing on this page clears a load path, and a licensed PE rules on all of it.

The honest secondary notes

The segment's own arithmetic

Aspect. Counted to the plate, option B is h/b 2.694 and option C is 1.553. Both sit inside the blocked wood-structural-panel limit of 3.5 that 4_ENG_SW's CODE BASIS block transcribes from SDPWS Table 4.3.4 — but only C clears the h/b 2.0 threshold at which SDPWS 4.3.4.2's Aspect Ratio Factor begins cutting capacity. B takes a factor of 0.913; C takes 1.000. Computed with the identical construction the workbook uses — verified by reproducing 9_TH_SW_CS's own published ARF (1.000 at aspect 1.9750) before printing anything.
Counted to the plate, not to the sloped ceiling. CHASE-SOUTH is counted to its full sloped ceiling (section 14.3); this page counts the extension to the plate, and says so. That reading is favorable on both currencies at once — a shorter wall has the better aspect ratio and hides less oak — which is exactly why it should be stated rather than assumed. Building it taller is a different option and a different page.
Where it bears. On the N double trimmer, the same doubled member the floor frames into — not on a beam. 2_LD_WIND section 14.6 flags the identical substrate question for CHASE-SOUTH's hold-downs at the S trimmer, and names it an unruled application for the PE. This one would be the same class of question at the other end of the cutout. Nothing here checks it.
It is dry-ish, not dry. The segment lives under the covered porch, on its north face weather-exposed. That is an assembly question (R-1a's exterior-class treatment for CHASE-SOUTH is the precedent), not a geometry one, and it is not priced here.

What else the same wall would do

It is the NW wind baffle the 2026-07-04 memo wished for. That memo calls NW “the site's hard corner” and asks for “a small wind baffle / half-jamb to knock down the worst drive-in while staying open”. An extension along the trimmer is exactly that — and note the memo's own words: small, and staying open. Option C is not small.
A port toward the oak is a real variant, and it is not drawn. An opening in the extension trades sightline back for Σb: any hole removes its own width from the counted length, and the segment is short enough that the aspect ratio of what remains is the binding worry, not the hole. ⛔ No port is sized, positioned, priced or drawn on this page — that is register 1.2's and 4_ENG_SW's work, and it needs the segmented-method layout the east wall got, not a sketch.
The alternative nobody has authored. The only documented way to stiffen this corner without a wall is the 2026-07-04 memo's “1 knee brace each way at the exposed NW post” — and that memo says in terms it is “belt-and-suspenders, not a portal frame”. ⛔ No document in this project designs, sizes or checks a porch portal frame, and none is drawn or dimensioned here. It is named as an engineering-hours item and left there.

⚠️ A standing document disagrees with the live workbook, and it is worth a dated look. The 2026-07-04 memo argued the NW notch was structurally comfortable partly because “~6 ft of N wall and ~6 ft of W wall” remain “intact as shear panels”. The live workbook rules the NORTH line STRUCTURALLY FREE with length 0 and counts nothing from it (2_LD_WIND section 11b, the six-wall table; no wall dead is charged for it either). The workbook is the later and more explicit statement, so it governs — but the memo's lateral argument rests on a premise the workbook has since withdrawn, and that is a fact the owner should have in hand when reading either. ⛔ Neither document was edited from here.

Deliberately NOT drawn

A recommendation. The owner deferred this to see it; presenting the trade is the whole job.
A portal frame — no member, no brace, no geometry, no capacity.
A port — no size, no position, no price, no header.
Any new wind arithmetic. Adding a third counted E–W line changes the tributary split and therefore the per-line demand on 9_TH_SW_S and 9_TH_SW_CS. That is calcs.xlsx's work and it is not done here; the V/Σb column is a direction-average comparison metric, not any line's design v.
Any child stature or eye height, and no glazing, plexi or ladder detail.
The door — register 3.2 is open and no leaf is sized here.
Anything east of X 66.5. The door wall exists; this page does not re-open it.

What this page would need in order to be decided

An owner ruling on how much of the oak the porch may lose — the table above is the currency, and it is a taste question, not an engineering one. Until it is made, every length here stays a candidate.
A ruled passage width at the trimmer line, since option C's length is set by it and no source states one. FIELD VERIFY at ladder procurement.
A calcs task if any length is adopted: Σb, a 9_TH_SW_* instance for the segment, hold-downs at both ends on the trimmer, the tributary re-split, and the torsion re-rank — the same bundle R7 and R10 each triggered.

STUDY — NOT PART OF THE DESIGN. Built on geometry.py DATUM v9 (2026-08-13) and a headless recalculation of calcs.xlsx, read by defined name. FROZEN 2026-08-13. Do not regenerate. A study records what was known when a question was framed; re-running it against a later datum produces a different document and destroys that record. • ⏸ OPEN — nothing here is selected. • This page is never to be measured from. • Planning-level only: a licensed PE and the AHJ rule on every structural number on it.

📁 archived · plate-corner-connections — click to view

⭐ RULED R28 (2026-08-14) — half-lap over the corner post LEAD (2-in tongues, pinned; a scoped, dated exception to the no-site-built-joinery position), butt + HTP37Z ALTERNATE, miter OUT, PE confirms both. R30 ruled the corner post SOLID rough 4×4; R32 then scheduled the corner ties (J13, datum v10).
⛔ FROZEN at its own datum · 2026-08-14 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-14-plate-corner-connections/

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V9 (2026-08-13) — ADOPTS NOTHING, SCHEDULES NOTHING — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. THE CORNER TIE IS UNSCHEDULED AND THIS PAGE DOES NOT SCHEDULE IT.

What is actually at the four corners of the top plate?

R27 closed the doubled-plate question on the ground that “the corner tie is hardware under every candidate”. That is right — and it makes the next question unavoidable, because the tie of record ties the plate to a member that is no longer primary. This page walks the plate ring corner by corner, from the sources, and draws what it finds. Three of the four corners have no scheduled tie and two have no stated member to tie to.

⛔ PROVENANCE — a STUDY, and studies do not get refreshed.

class: STUDY (illustrative). Opposite staleness rule to the construction set — see studies/README.md and docs/decisions/2026-08-09-study-vs-set.md.
frozen on: DATUM v9 (2026-08-13) — read off geometry.py’s own banner at generation, and asserted (DATUM_LIVE = v9). Re-running this generator on a later datum produces a different document and destroys the record.
generated: 2026-08-14
ruling behind it: R27 (owner, 2026-08-14): R5 RE-AFFIRMED — the E & W top plates are a SINGLE ROUGH TRUE 4×4. Owner: “ok you convinced me double out single 4x4 rough in.” Memo docs/decisions/2026-08-14-top-plate-corner-question.md. The reversal-to-doubled-2×4 question is CLOSED; this page is the follow-on work that memo commissions.
and R28: R28 (owner, 2026-08-14, mid-study): the corner detail is RULED in FORM. Owner: “c mitre out; i think b sounds like lead followed by a with Simpson Strong-Tie HTP37Z 3″ x 7″ Heavy Tie Plate … for tie; draw both in study.” → miter OUT; half-lap LEAD; butt + HTP37Z ALTERNATE; both drawn, PE confirms both. Same memo, addendum.
and R29: R29 (owner, 2026-08-14, mid-study): the C/H 4×12 gable header alternate is RETIRED. Owner: “also retire 4x12 header; 4 ridge posts is locked unless we encounter objections.” The four-ridge-posts-on-four-4×10-crosses scheme is now the sole scheme of record, still pending its two standing gates (PE review · Treehouse Supplies). Retire ≠ delete. ⛔ This page describes the primary scheme only; C/H appears as a retired status note and nowhere else. ⛔ R29’s propagation is a queued work package and is not this study’s scope — geometry.py still carries the pre-R29 statuses, and the generator reads the live value so the page cannot drift from the datum it is frozen on.
⚠ scoped exception: The half-lap is a knowing, dated, scoped exception to this project’s standing no-site-built-joinery position (owner, R28, 2026-08-14): four simple square-cut laps over bearings, and nothing else. The position still bars compound-angle and load-path-critical joinery. Recorded here because a reader three months from now will otherwise read the lap as a drift.
premise, asserted: PLATE_W = PLATE_DEPTH = 4 in, PLATE_STEP = 0 (flush). The generator refuses to build if any of those move.
adopts: Nothing. Both ties in Figure 3 are options. The proposed J13 row is a proposal; J-rows enter the schedule in a Fable datum pass (J11/J12 did, datum v6) and a study may not execute one.

① The plate ring, and the six joints in it

Every perpendicular member at Z 66 on file is 4 in wide — R21’s door-wall plate, the eave plates themselves, and (until R29 retired it) the C/H header. Which is why the tie geometry in Figures 3 and 4 can be drawn honestly while the member’s identity stays unstated. The landing is 4 in whatever the answer turns out to be — and R29 makes that answer overdue rather than merely open, because the retired scheme was the only one that ever named a member here.

② The four corners, established from the sources

	where	what stands there at Z 66	what the plate bears on	tie SCHEDULED today
C1	WEST plate sta 16, X 0	N gable line, X 0→108. ⛔ No top member is stated. The C/H alternate put a rough 4×12 header here with its TOP at the plate line — and R29 (owner, 2026-08-14) RETIRES that scheme, so the answer it used to give is gone with it.	The NORTH end of register 1.2e's porch west wall — a FRAMED BEARING WALL, sta 16→43.5, full-height studs, two open ports. ⛔ No source names the post at its north end.	⛔ UNSCHEDULED J8 is the only tie ever written for this joint and it is re-statused NOT PRIMARY (C/H) — it ties plate → HEADER.
C2	EAST plate sta 16, X 108	Same N gable line. Same silence.	⛔ NOTHING IS RULED. Register 1.2e rules the WEST porch segment only; the wall inventory's EAST line starts at station 43.5. The east plate's northern 27.5 in has no stated support and no stated end post.	⛔ UNSCHEDULED There is not yet a member for a tie to land on, on either leg.
C3	WEST plate sta 138, X 0	SOUTH wall, station 138, X 0→108, the primary E–W shear line (9.0 ft). ⛔ Its TOP MEMBER at Z 66 is not stated — plate-at-66 or studs to the rake is the same unanswered question R20 gate item 5 asks about the north wall, and R29 removes the retired header that used to stand in for it.	The SOUTH end of the WEST wall (register 1.4, ruled 2026-08-13 R7: full-height framed + sheathed, sta 69.75→138, no openings). ⛔ Corner post OPEN — R29 orphaned the ruling.	⛔ UNSCHEDULED J8's member is not primary; nothing replaced it.
C4	EAST plate sta 138, X 108	Same SOUTH wall, same unstated top member. The R18 SE-3 egress window is sited HARD EAST in this wall.	The SOUTH end of the EAST wall (bearing, sta 43.5→138). ⛔ Corner post OPEN, as C3.	⛔ UNSCHEDULED J8's member is not primary; nothing replaced it.

And the R21 door-wall plate’s two ends

	where	what it meets	what it bears on	tie SCHEDULED today
T1	sta 43.5 X 66.5	R21 door-wall plate, WEST end. It arrives at the N-mid RIDGE POST, which RAISE 7 establishes is CONTINUOUS to the ridge soffit (107.635 in) and carries 806.1 lb.	⛔ It cannot bear on top of anything — the post runs past Z 66. It must frame into the post's SIDE.	⛔ UNRULED APPLICATION. No connector for a side-framed plate → continuous post is seeded in docs/components/. RAISE 7 names the gate; never-invent forbids picking one here.
T2	sta 43.5 X 108	R21 door-wall plate, EAST end, stated at X 108 — which is the EAST WALL LINE, where the east eave plate already occupies X 104→108 as one uncut 122-in stick.	⛔ Over-committed as written: two members cannot share X 104→108 at Z 66. If the east plate stays continuous (it must — the zero-splice premise), the door plate BUTTS its west face at X 104 and the member is 33.5 in, not 41.5.	⛔ UNSCHEDULED, and it is a T-INTERSECTION, which Table R602.3.2 ties on the same row as a corner.

⭐ Why every “tie” column says UNSCHEDULED. J8 — “side-wall top plates → header (corner)” — is the only tie ever written for a wall corner in this project, and its status in geometry.CONNECTIONS is NOT PRIMARY (C/H) as geometry.py still reads it at datum v9 (datum v6, 2026-08-08: the joints that name the gable header follow the header into the alternate) — and R29 now RETIRES that scheme outright, so J8 retires with it. ⚠ That re-status is queued propagation, not this page’s work; the generator asserts the live datum value so the two cannot silently disagree. The locked four-post scheme’s own joints are J11 (ridge → post) and J12 (post base → bearing chain), both PE TO SPECIFY. Neither is a wall corner. The BACKLOG has carried this since the day of the demotion: “the four-post scheme’s gable-corner tie has NO source … deliberately NOT invented”.

③ The corners in section — and the post nobody has settled

Two live sources, one corner post, and the reason one of them gave has evaporated.

materials_checklist.md: “Solid gable corner posts — rough 4×4×8 × 4. ⛔ Do NOT let a BOM cleanup revert these to 4½″ built-up 3-stud corners — they are solid because of the strap schedule: a built-up post steps every strap ¼″ per side.”
phase2_year2_cabin.md §2.2, the wall-framing basis: “Corners = built-up stud posts over the beam line.”
geometry.MEMBERS["corner_post"]: solid rough 4×4 — and annotated ⛔ NOT PRIMARY (datum v6), “the header alternate’s bearing posts”.

⭐⭐ R29 turned this from a disagreement into an ORPHAN. The solid 4×4 was adopted 2026-08-05b as part of the C/H stack (“one width through the whole gable stack”), and the checklist’s stated reason is the strap schedule — J1–J4, every one of them a C/H joint. Retiring C/H retires the reason, and what is left standing is the 2026-07-22 built-up-corner decision and the general wall-framing basis. On a built-up 3-stud corner at 4.5 in the v9 plate steps 0.25 in per side — the same quarter inch datum v9 rule C declares dissolved, reappearing with its sign reversed (post wider than plate, not narrower). ⭐ And R28’s half-lap wants the solid 4×4: a 4-in plate laps flush over a true 4×4; over a 4.5-in built-up it leaves a half-inch step and seats the lap on a layered top. ⛔ OPEN — owner ruling owed. This page draws neither as resolved.

④ ⭐ The corner detail, as R28 ruled it

⭐ R28 (owner, 2026-08-14) — the FORM is ruled, the SPEC is not. Three options were put up: (a) run-through + butt with a flat steel tie, (b) half-lap over the corner post, (c) 45° miter.

option	status	why it matters here
(b) HALF-LAP over the post	⭐ LEAD — owner, R28	Each plate keeps a 2-in tongue; interlocked; the joint bears fully on the post; pinned through the lap into the post.
(a) BUTT + HTP37Z	ALTERNATE — owner, R28	Owner-specified part. 3″×7″ Heavy Tie Plate, ZMAX, flat on the top face across the joint.
belt-and-suspenders vertical corner straps	⭐ PROPOSED — owner, R28 follow-on	One flat strap on each outer corner face, post face up across the outside of the lapped plates. 2 per corner, 8 total. Completes the corner as an uplift node and backs the lap pin. ⛔ Part not final — MSTA12Z-class candidate, registry check + PE.
~~(c) 45° MITER~~	⛔ OUT — owner, R28	~~Kept as a struck row only, per the register’s nothing-is-deleted rule.~~

⛔ R28 rules the form. It does not schedule a joint, size a pin, or state a capacity — and neither does this page.

The lap square is the post’s own footprint, and the eave plate must take the upper tongue. 4×4 in, because POST_W = PLATE_W = 4. And the handedness is derived, not chosen: the rafter birdsmouth seats on the plate top within 2.5 in of the outer face — inside that square. A lower-tongue eave plate would remove the seat at exactly the corner the end rafter lands on. The generator refuses to print the argument if that ever stops being true.

⛔ Three things the half-lap carries that a steel tie does not.

It removes 50 % of each member at the corner (2 of 4 in). The plate-notching provision this project could find is R602.6.1, which addresses cuts for piping and ductwork and calls for a tie above 50 %. Nothing on file states whether it reaches a structural half-lap, and this page does not decide it. PE / AHJ.
It is not a tie listed in Table R602.3.2. R602.3.2’s exception permits the single plate where tied per that table. A lap is not on it. So the half-lap on its own leaves the prescriptive path and becomes an engineered alternative under R104.11 — the alternate (a) stays on it.
⭐ They are not mutually exclusive. Because the eave plate takes the upper tongue, the corner’s top face is continuous — so a flat plate can ride over a lapped corner. Whether both are wanted is the PE’s call; the geometry does not forbid it.

⛔ The pin is the one thing no source states. The owner asked for it from established practice or manufacturer literature. What is on file: nothing describing a pin for a half-lapped top plate. What can be said without inventing is geometry — the lap is 4 in of wood, so a screw of length L leaves L − 4 in the post, and this project’s own SDWS22800DB (8 in Strong-Drive SDWS TIMBER, IAPMO-UES ER-192) leaves exactly 4 in. ⛔ It is named as a candidate on length alone. No count, no edge distance, no capacity, no pattern. And the registry rule bites: SDWS22800DB is APPROVED at guard-post-base and nowhere else — this is a new application and needs its own dated row. PE TO SPECIFY.

⑤ The HTP37Z, from the manufacturer’s own table

⭐ Definitive data — transcribed verbatim from Simpson Strong-Tie Wood Construction Connectors catalog C-C-CAN2022 p.285, 'Strap Ties (cont.)' table, HTP37Z row. ⚠ That is the Canadian catalog and its numbers are factored resistances to CSA O86-14 (LSD), not ASD allowable loads; K_D is the CSA duration-of-load factor. The US allowable-load row is NOT on file anywhere in this repo, and the ESR on file (ESR-2105) does not cover the HTP family — checked.

property	published value	note
gauge	16 ga	= 0.065 in in ZMAX (G185), or 0.06 in uncoated — catalog p.24, US Standard Steel Gauge Equivalents
dimensions	W 3 × L 7 in	3.5 in each side of a centered joint
fasteners (total)	(20) 10d x 1 1/2" (0.148" dia. x 1 1/2" long)	10 per member — catalog note 2: “Use half of the nails in each member being connected to achieve the listed resistances.” (the same half-the-fasteners rule ESR-2105 imposes on the MSTA straps, and the rule that killed the MSTA24 at J1)
factored resistance D.Fir–L	1,610 lb (K_D 1.00) · 1,850 lb (K_D 1.15)	⛔ LSD, not ASD. These are not comparable to the 285.7 lb ASD collector demand, and this page does not compare them. Mixing strength and ASD bases has already produced one false alarm on this project.
factored resistance S-P-F	1,270 lb (K_D 1.00) · 1,460 lb (K_D 1.15)

What CAN be said, and it is what R28 said: the HTP37Z exceeds the prescriptive minimum — in steel and in size. 0.065 in against Table R602.3.2’s 0.036 in is 1.81×; 7 in against 6 in is 1.17×. ⭐ And the code’s 0.036 in is 20-gauge uncoated steel on Simpson’s own equivalents table — so the comparison is like for like, and it is a steel comparison, not a load one. No basis conversion is needed to make it, which is exactly why this page makes it and makes no other.

⑥ Why the tie had to be re-chosen at all

The prescriptive code plate fits the member. The scheduled strap does not. Option (i)’s 3-in leg lands inside 4 in with 1 in to spare. Option (ii) is scheduled 7 fasteners over 9 in per member and there are only 4 in of member — roughly 3–4 holes at the NOMINAL pitch. ⚠ FIELD VERIFY: STRAP_HOLE_PITCH_NOMINAL is a declared range, not a published pitch, and 2.6a’s own standing rule is to lay the real strap against the real member before ordering.

⛔ What this page does NOT do with that. It does not state an allowable for a reduced fastener count. ESR-2105’s 1,315 lb is a listed value for the listed fastening; converting a hole count into a capacity is a PE call under R104.11, and doing the division here would be exactly the kind of authoritative-looking guess the never-invent rule exists to stop. What the page states is a count, which is geometry.

⑦ The code panel — 2021 IRC R602.3.2

⚠ Transcribed, not read from an original. No IRC page is on file in this repo. Everything in this panel comes from docs/decisions/2026-08-14-top-plate-corner-question.md §1, which consolidates a sourced code-research pass. Staple the real page into docs/product_specs/ before any of it is called independently verified — the same declaration the component registry makes about its own citation gaps.

the exception	what it requires	this building
R602.3.2 Exception	A single top plate is permitted where the plate is tied at joints, corners, intersections and in-line splices per Table R602.3.2.	⭐ The construction R5/R27 ruled is prescriptively sanctioned — it is not an engineered departure.
Table R602.3.2 SDC A/B row	3″ × 6″ × 0.036″ galvanised plate, (6) 8d box nails each side, at corners and intersections.	The site is SDC A/B. A Simpson TP37 matches the spec — but see the registry warning below.
in-line splices	12″ plate, 12 nails.	⭐ Moot. Each 122-in run is one 12-ft stick, zero splices — which is also why the zero-splice premise must survive any change at T2.
condition 2	Rafters/joists centered over studs within 1 in.	The design stack-frames. ⚠ A standing requirement, forever, not a one-time check — it is the discipline a single plate buys its simplicity with.
condition 3	The plate may be omitted over headers where the header is “adequately tied to adjacent wall sections in accordance with Table R602.3.2”.	⭐ This is the code’s own answer to the C/H gable header: a steel tie, never a lap — which is precisely where 2.6a §6.5 landed independently in 2026-08-05b.
jurisdiction	The local building code enforces the 2021 IRC for new permit applications.	⚠ the local AHJ edition is confirmed at the memo’s level of research only; the J10 entry records that an edition question has bitten this project before.

⛔ THE COMPONENT-REGISTRY TRAP, NAMED. The TP37 is already on this project — two of them, tying the Year-1 west drop-beam splice (3.125×7 in, 20 ga, 32 holes; 1.6_drop_beams.md). That approval is for that application and does not transfer here. A component may be approved in one application and rejected in another — and the TP37 is not in docs/components/ at all (8 of ~18 seeded), so its absence means not yet seeded, never no ruling exists. Two things travel with it that a corner would have to answer: (a) coating — the TP37 is G90 only, there is no ZMAX SKU, against a project coating rule of ZMAX + HDG/SD on green wet-service stock; (b) fasteners — its listed fastening is 0.131×1½ nails or #9×1½ screws, while Table R602.3.2 calls for 8d box nails. Those are not the same fastener, and nothing on file reconciles them.

⑧ ⛔ OPEN ITEMS — raised, not resolved

A study reports; the owner rules; the owning document is fixed in its own commit (discrepancy protocol). Nothing below was changed anywhere.

⛔⛔ THE CORNER POST IS OPEN — OWNER RULING OWED, and it is now the first thing in the way. The solid rough 4×4 corner posts were adopted 2026-08-05b as part of the C/H stack; before that, all corners were built-up 3-stud dimensional (4.5 in) per the 2026-07-22 decision. R29 retires C/H and silently orphans the adoption — and the checklist’s own stated reason for going solid (the J1–J4 strap schedule) is retired with it. ⭐ R28’s half-lap wants the solid 4×4: a 4-in plate laps flush over a true 4×4, but over 4.5 in it leaves 0.25 in of step per side and seats the lap on a layered top. Owner ruling owed on the corner-post spec independent of the retired scheme. Nothing is drawn as resolved either way.
⭐⭐ The wall corner tie is UNSCHEDULED — R28 rules the FORM but no J-row carries it, and R29 makes the gap permanent. Before R29 one could at least say “if C/H is ever built, J8 returns”. With the alternate retired there is no scheme on file in which a wall-corner tie is scheduled at all. J8 names the header and is NOT PRIMARY; J11/J12 are the ridge post’s joints, not the wall’s. Proposed as a NEW row J13 — “eave top plate → perpendicular top member at the wall corner and at intersections”, status PE TO SPECIFY, six locations (C1–C4 plus T1/T2), carrying R28’s pairing verbatim: half-lap over the post PRIMARY (with the proposed vertical corner straps), butt + HTP37Z ALTERNATE, PE to confirm BOTH — with the pin spec, the strap part and the HTP37Z’s ASD basis named as the three things the row cannot fill in. ⛔ The number is a proposal only. J11 and J12 entered geometry.CONNECTIONS in a Fable datum pass (datum v6); J13 must arrive the same way, and this page may not do it.
⭐ The belt-and-suspenders straps are PROPOSED, and their part is not final. R28 follow-on: one vertical flat strap on each outer corner face, post face up across the outside of the lapped plates — 2 per corner, 8 total. ⛔ Not an HTP37Z: a tie plate is a coplanar-splice part and this is a corner edge. The natural family is MSTA12Z-class — and note that MSTA12Z’s only cabin ruling on file (J4) is a C/H joint that R29 retires, so this is a new application with no live precedent. Registry check + PE confirm; never inferred. ⛔ And the geometry has a real edge: set flush with the plate top the strap gets 4 in of plate and 8 in of post — but within the lap square that 4 in is 2 in of the eave plate plus 2 in of the perpendicular member, so a two-member listing is being asked to serve a three-member joint. At the NOMINAL hole pitch that is about 2 holes per tongue — ⚠ FIELD VERIFY, and a PE question, not an arithmetic one. Sheathing normally passes over a flat strap; strap-over-sheathing carries reduced values — cite Simpson or tag PE before either is assumed.
⛔ The half-lap’s PIN is unsourced — no document on file states a pin for a half-lapped top plate. SDWS22800DB is named as a candidate on length (8 in through 4 in of lap leaves 4 in in the post) and on nothing else. Its registry ruling is guard-post-base only. PE TO SPECIFY; new registry row owed.
⚠ The half-lap’s 50 % notch has no provision on file that reaches it. R602.6.1 addresses plate cuts for piping/ductwork. Whether a structural half-lap at a bearing point falls under it, and whether a tie is then required anyway, is a PE / AHJ question. Related: the lap alone is not a Table R602.3.2 tie, so it is an engineered alternative (R104.11), not the prescriptive path — the ALTERNATE (a) is.
⛔ The HTP37Z’s US ASD allowable is NOT on file, so no margin against the 285.7 lb demand can be stated. What is on file, and transcribed above, is the Canadian catalog’s factored resistances (CSA O86 LSD). ESR-2105 — the one Simpson evaluation report in docs/product_specs/ — does not cover the HTP family (checked page by page). Staple the US C-C allowable-load page into docs/product_specs/ before any margin is quoted anywhere.
⚠ HTP37Z is not in the component registry — absence means not yet seeded, never no ruling exists. A ready-to-move draft doc is left beside this study as HTP37Z.seed.md; it was deliberately NOT written into docs/components/, which is not this session’s declared domain. Owner: move it and re-run tools/registry/build_index.py. ⚠ Note it is already keyed in the BOM (Multi-Use Materials A28), so the part is priced before it is ruled.
⭐ The top member at BOTH gable lines is stated by no source under the primary scheme. Station 138 (the SOUTH wall, the primary E–W shear line) and station 16. The C/H alternate answered it with a rough 4×12 header whose TOP was the plate line; that member is demoted and nothing replaced it. This is the same question R20’s gate item 5 asks about the NORTH wall — a wall, not a number: does the gable end have a top plate at Z 66, or do its studs run to the rake? It is now owed on three wall lines, not one.
⛔ The EAST porch segment, station 16→43.5, has nothing ruled at all. Register 1.2e rules the WEST porch segment a framed bearing wall with two ports; the wall inventory’s EAST line starts at station 43.5. So 27.5 in of east plate has no stated support, no stated end post, and corner C2 has neither leg of a tie to land on. ⚠ It also means the two plates are not symmetric in support, which 9_TH_EAVE_W’s single “west eave run” framing does not currently reflect.
The corner post TYPE is disputed and the reason for the ruling has evaporated — solid rough 4×4 (checklist, justified by the NOT PRIMARY strap schedule) vs built-up 3-stud (wall-framing basis §2.2). At 4.5 in the plate steps 0.25 in per side, which is the step datum v9 rule C says dissolves. Owner/PE.
⛔ R21’s door-wall plate is over-committed at both ends. West (T1): its stated start X 66.5 is the west face of the N-mid ridge post (66.5→70.5), and RAISE 7 establishes that post is continuous to the ridge soffit (107.635 in) — so nothing can bear on top of it and the plate must frame into its side. East (T2): its stated end X 108 is the east wall line, where the east eave plate already occupies X 104→108 as one uncut stick. The real clear run between the two is 70.5→104 = 33.5 in — which reproduces R21’s own “33.5 in clear” exactly, so the arithmetic is consistent and only the member length (41.5 in as written) and the two end conditions are unstated. Both ends are side-framed connections. RAISE 7 names the west one and calls it an unruled application; the east one is named nowhere.
⚠ Two live figures for one plate run. R24 / BOM G129 say 122 in (station 16→138); 2.6a §6.5 ① says 121.25 in (10'-1 1/4"). 0.75 in apart, no consequence to the buy (12 ft either way) — but two documents state different values for one quantity. R27 §5 item 4 already raises it; repeated here because a corner study is where the ends of that run get drawn.
⚠ The collector demand’s own note is stale, twice. Live: 285.7 lb (MEMO 2026-08-14-top-plate-corner-question §4b/§5 · CALC: 2_LD_WIND cell G488 (live)). The tab’s note still reads ~429 lb and still calls the member a doubled dimensional 2×4; 2.6a §6.4’s J8 row says the same. The held PE brief quotes the margin as 3.1×. Propagation fixes — Domain B and Domain C, not this page.
⚠ The stud pre-cut nobody carries. Under the v9 4-in plate the E/W-wall studs are 60.5 in (66 − 4 − 1.5); Cabin (Year 2) C70/C71 still hold 61.5, and R21 extends the 4-in condition to the north door wall. Raised in R27 §5 item 3; drawn here because Figure 2 dimensions it.
⚠ TP37: right part, wrong ledger. Not in the component registry; approved on this project only at the Year-1 beam splice; G90 only against the ZMAX coating rule; its listed fastener is not the code table’s 8d box. Seed it before it is specified anywhere near a corner.
⚠ The R21 plate’s N–S faces are unstated — Figure 1 draws it centered on station 43.5 as a drawing convention and says so on its face. And raise 7’s side-framing connector remains the gate on R21 as a whole (owner-recorded, 2026-08-14).

Deliberately NOT drawn

A recommendation · a pin spec, count, pattern or edge distance · a fastener schedule · any allowable for a reduced fastener count · any ASD figure for the HTP37Z · a capacity verdict · a new J-row in geometry.CONNECTIONS · a component-registry ruling · a strap part, length or nailing · any of R29’s propagation (2.6a, the BOM, the set faces, the registry rows) · a size, species or grade for any member the sources leave unstated · a top plate on either gable wall · a corner post type · a wall on the east porch segment · a price · any new wind, uplift or shear arithmetic. R28 ruled the corner’s FORM; this page draws that form and stops there.

What the generator refuses to build

the datum gate — geometry.py’s banner must still read DATUM v9, read off the banner itself
the premise gate — PLATE_W and PLATE_DEPTH must still be 4, PLATE_STEP still zero
the demotion gate — J8’s status must still be NOT PRIMARY (C/H), and no plate joint may have become primary; this whole page is downstream of that and may not quietly outlive it
J13 must be the next free row and must not already exist
the “4 in wherever it lands” finding is asserted against HEADER_W, PLATE_W and POST_W together — if one moves, the tie geometry is refused
⭐ R28’s half-lap must actually be a HALF lap — the 2-in tongue must be exactly half PLATE_DEPTH, the lap square must be the post’s own footprint, and the rafter seat must fall inside that square, or the handedness argument does not hold and does not print
⭐ R28’s “exceeds the prescriptive minimum” must reproduce from the catalog’s own gauge table — and the assertion that 0.036 in is 20-gauge uncoated steel is checked against that table, not assumed
the HTP37Z’s 20 fasteners must split evenly between the two members, per the catalog’s own note 2
R21’s 41.5 minus the derived 33.5 must be exactly the two 4-in members it spans between, or the over-commitment finding is wrong and the page does not print it
the strap’s scheduled half must be longer than the member it lands on, or the counted-nailing finding has no basis
the four ridge reactions must still sum to the published total; both eave seats must still land on the plate
handedness — Figure 1 is a plan from above with north up, so east plots RIGHT and station increases DOWN; Figure 2 is labelled LOOKING NORTH, so RIGHT_HAND puts WEST on the right and every mapper must run X to the LEFT; and up must be up
every SVG label measured against its own viewBox, horizontally and vertically, on a deliberately generous character-width table
a forbidden-claim scan (no adoption, no schedule, no derated allowable) and a required-phrase scan (FIELD VERIFY, UNSCHEDULED, PLANNING — NOT FOR CONSTRUCTION, the datum, R27)

Study generated 2026-08-14 from treehouse/tools/draw/geometry.py at DATUM v9 (2026-08-13) (banner read live: v9). Walls and rulings from the interiors register, 2.6a, the build guide and the dated memos, each cited where used. STUDY — NOT PART OF THE DESIGN. PLANNING — NOT FOR CONSTRUCTION. The generator beside this file is provenance, not a build step: it proves every number came from the datum rather than from a person’s estimate. ⛔ Do not re-run it.

📁 archived · plate-ring-layout — click to view

⭐⭐ RULED R32 (2026-08-14) — the six-post four-stick RING + adoption package ADOPTED (datum v10); R34 (2026-08-15) dissolved this study's two-inch registration binary — posts center on the SQUARE-frame nailers, fully bearing, neither reading the right frame. R35c (2026-08-15) later deleted the two station-16 segments for the cathedral porch.
⛔ FROZEN at its own datum · 2026-08-14 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-14-plate-ring-layout/

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V9 (2026-08-13) — ADOPTS NOTHING, SCHEDULES NOTHING — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. THE RING IS PROPOSED — THE OWNER IS DECIDING, AND NOTHING ON THIS PAGE IS RULED.

The proposed top-plate ring, drawn on the datum

Four single rough true 4×4 sticks at Z 66, closing a ring around the station 16–138 envelope: the two eave plates that already exist, plus a new plate on each gable line. Six solid posts carry it, two of which frame the west ladder entry. The owner asked to “visualize or more detail” before ruling. Here is the ring as the sources actually place it — including the two-inch registration clash it walks into at every single post, which no source has had to answer before because no member has ever been asked to land on a gable line.

⛔ PROVENANCE — a STUDY, and studies do not get refreshed.

class: STUDY (illustrative). Opposite staleness rule to the construction set — see studies/README.md and docs/decisions/2026-08-09-study-vs-set.md. Do not re-run this generator on a later datum.
frozen on: DATUM v9 (2026-08-13) — read off geometry.py’s own banner at generation and asserted (DATUM_LIVE = v9).
generated: 2026-08-14
status: ⛔ PROPOSED. Not ruled, not adopted, not drawn into the datum. If it is ever ruled it is a datum change (new members) → a Fable v10 pass, natural to merge with the queued rule-D pass and R29’s propagation.
the proposal: Owner, 2026-08-14: a closed ring of four 4×4×12 ft sticks on six posts, the north plate doing “double duty as door header” over the porch face. Recorded in docs/decisions/2026-08-14-top-plate-corner-question.md under PROPOSED, with R31 locating posts 4 and 6.
the amendment: Owner, same day: “n & s top plates can’t be continuous bc the n & s ridge posts extends from floor to ridge beam.” ⭐ Source-confirmed here: RIDGE_POST_FLOOR_TO_SOFFIT = 8'-11 5/8" holds at all four stations (2.6a, datum v6 rule D), so each gable plate is two segments side-framing into the post faces. The generator refuses to build if that constant ever stops clearing the plate.
rulings behind it: R27 (the eave plates stay a single rough 4×4) · R28 (corner detail: half-lap LEAD, butt + HTP37Z ALTERNATE) · R29 (the C/H 4×12 header RETIRED) · R30 (corner posts SOLID rough 4×4) · R31 (posts 4 and 6 located; they frame the west entry). All owner, all 2026-08-14, all in the one memo.
sibling study: The FROZEN 2026-08-14-plate-corner-connections established the corner conditions and the lap geometry. ⛔ It was not modified and not re-run — it is quoted.
adopts: Nothing. No connector is selected, no member is sized, no capacity is claimed and no J-row is written.

① The ring, as the sources place it

plate	status	line	length	stick
WEST eave	✅ EXISTS — R5 / R27	X 0→4, sta 16→138	10'-2"	one uncut 12-ft stick
EAST eave	✅ EXISTS — R5 / R27	X 104→108, sta 16→138	10'-2"	one uncut 12-ft stick
SOUTH gable	⭐ NEW — PROPOSED	sta 138 line, two segments: X 0→66.5 and X 70.5→108	5'-6 1/2" + 3'-1 1/2" = 8'-8"	one 12-ft stick, offcut 3'-4"
NORTH gable	⭐ NEW — PROPOSED	sta 16 line, two segments, same split	5'-6 1/2" + 3'-1 1/2" = 8'-8"	one 12-ft stick, offcut 3'-4"

The segment split is not a choice: 66.5 + 4 + 37.5 = the 108-in wall span exactly, because the ridge post occupies X 66.5→70.5 on the datum. Kerf is not counted in the offcut.

Each gable line still cuts from ONE 12-ft stick — with 3'-4" left over. 5'-6 1/2" + 3'-1 1/2" = 8'-8" out of 12'-0". The two-segment amendment costs nothing in material; four sticks still buy the whole ring.

② The six posts, and what each one lands on

post	footprint at Z 66	bears on	source
1 · NE	X 104→108 sta 16→20	doubled 2×8 N gable nailer over the flush 4×10 N cross	owner R31 table row 1 · MEMBERS['nailer'] / ['cross']
2 · SE	X 104→108 sta 134→138	doubled 2×8 S gable nailer over the Year-1 4×10 S cross	owner R31 table row 2
3 · SW	X 0→4 sta 134→138	doubled 2×8 S gable nailer over the Year-1 4×10 S cross	owner R31 table row 3
4 · W-ENTRY N JAMB	X 0→4 sta 39.5→43.5	the N DOUBLE TRIMMER (station 43.5 C-C, band 41.5→45.5)	owner R31: “yes n double trimmer”
5 · NW	X 0→4 sta 16→20	doubled 2×8 N gable nailer over the flush 4×10 N cross	owner R31 table row 5
6 · W-ENTRY S JAMB	X 0→4 sta 69.75→73.75	⛔ OPEN — see the bearing note	owner R31: “n end of nook eg s of ladder siting”

All six are R30’s SOLID rough true 4×4. ⚠ geometry.MEMBERS["corner_post"] still reads “SOLID rough 4×4 hemlock (TRUE 4.0 in) — ⛔ NOT PRIMARY” — the NOT-PRIMARY annotation belongs to the C/H stack R29 retired, and R30 re-ruled the post independent of that scheme. The generator reads the live annotation so this note cannot silently outlive the queued propagation.

⛔ OPEN — the bearing under post 6. Post 6 sits on the west wall line at station 69.75. What the sources give, derived, without inventing a detail:

The west beam is continuous beneath the west wall line (Year-1, no splice) — but its as-driven ends are PILE_SW (0, 0) and PILE_NW (-2.625, 121.25), so the beam is tilted off the idealised X 0 line. ⚠ FIELD VERIFY.
No joist lands under the post. Its footprint sta 69.75→73.75 falls inside the bay between joists J5 (sta 64.5 C-C) and J6 (sta 75 C-C) — 5.25 in and 1.25 in clear of those centerlines.
The nearest support under the beam going south is the planned Year-2 west-mid TAB at station ≈ 84.2 off the north rim (W_YR2_MID_STATION_PLANNED_NOMINAL 53.8 up-string from the SW post) — 14.45 in south of post 6. ⚠ That support is NOMINAL, root-blocked at probe, FIELD VERIFY, and it is PE + arborist gated.

⛔ No source states blocking, a squash block, or any bearing detail at station 69.75. Never-invent: this page draws the post and leaves the detail open. Owner / PE.

③ The west elevation — the porch, the entry, and the duck

The ring’s two jamb posts land exactly where the eave-plate calc already assumed bearing. The RULED unsupported west-eave run is 26.25 in = 2'-2 1/4" — station 43.5 (the porch wall’s south end) to station 69.75 (the west wall’s north end), util 0.270, shear-governed. That is the same opening posts 4 and 6 jamb. The ring does not lengthen any eave span; it gives the two ends that the calc already treats as supports a named post. [MEMO 2026-08-14 R27 §2 · CALC: calcs.xlsx '9_TH_EAVE_W' §6 (inversion), R25 re-base]

Duck clearance, both places the owner named. The plate underside is Z 66 − 4 = 62 in = 5'-2", and that is the head height at every point the ring crosses: the porch landing, the west ladder entry and the open north face alike. A 6-ft adult ducks at all three. ⭐ Outside the wall the ladder arrival is lower still — 4′-11 5/16″ under the datum-v8 tapered eave tail (register 0.4a, ruled; quoted here, not re-derived) — so stepping in through the entry is a gain of about 2¾ in, not a loss.

④ The north gable line — two segments, and a wall nobody has ruled

⭐ This is the drawing that makes the ring worth ruling — and the one that shows what it does NOT settle.

What it settles: a gable line with a plate at Z 66 is a flat-topped wall. That is the “a wall, not a number” question R20 gate item 5 asks about the north gable, and the ring answers it in the affirmative on both gable lines at once — which is exactly why the answer must come from the owner and not from a study.
⛔ What it does not settle: what is framed above the plate. No source states the gable-end framing on any gable line — not studs to the rake, not sheathing, not an infill. It is drawn here as an open triangle, and it stays open.
⛔ The cascade, stated on the face: the north-gable answer is live in the R20-held door-height discrepancy (docs/decisions/2026-08-14-door-height-discrepancy-and-pe-hold.md) — 9_TH_HDR_DOOR models a flat top, the 2026-08-08 door memo models a raking one, and the difference is worth 17¼ in of door leaf. Ruling the ring rules that too. Nothing on either side is corrected here.

⛔ OPEN — the north plate does BEAM duty over the open porch face, and nobody has checked it. The owner named it (“double duty as door header”). Its two clear spans, derived: 5'-2 1/2" west of the ridge post (post 5's face → post face) and 2'-9 1/2" east of it. ⭐ That east figure reproduces R21’s own 33.5 in exactly — same two members, same gap — which is the arithmetic behind the “one detail, six joints” claim below.

⚠ For scale only, and it does not transfer: the eave plate’s own governing clear span at utilization 1.0 is 68.88 in, deflection-governed (L/360) [MEMO 2026-08-14 R27 §2 · CALC: calcs.xlsx '9_TH_EAVE_W' §6 (inversion), R25 re-base]. The north plate carries a different load case — whatever the gable framing above it turns out to be, which is itself open, plus any door-header duty. ⛔ A member check is owed and it is Domain B’s to run: a new row alongside 9_TH_EAVE_W, on the ruled gable framing, not on this page’s arithmetic. The south plate needs no such check — it sits on the 108-in SOUTH wall for its whole length.

⑤ One detail, up to six joints

The side-framed plate end is the ring’s single biggest ask — and it is not new. Because the ridge posts run continuous, a plate meeting one cannot bear on top of it: the end grain frames into the post’s face. R21’s door-wall plate already has that condition at both of its ends, and raise 7 names the connector as unruled. The ring adds four more instances of the same detail. ⭐ That is a cleaner PE ask than six one-offs: one detail to rule, up to six places to use it.

joint	where	ruling status today
R21 door plate, west end	sta 43.5 · X 66.5	raise 7 names it — UNRULED
R21 door plate, east end	sta 43.5 · X 104	named nowhere — frozen sibling study, open item 5
N ring plate, west segment	sta 16 · X 66.5	NEW — this proposal
N ring plate, east segment	sta 16 · X 70.5	NEW — this proposal
S ring plate, west segment	sta 138 · X 66.5	NEW — this proposal
S ring plate, east segment	sta 138 · X 70.5	NEW — this proposal

⛔ OPEN — the connector itself. Nothing in docs/components/ covers a plate-end-to-post-face application — checked against the index (8 components seeded; absence means “not yet seeded”, never “no ruling”). No manufacturer literature on file addresses it, and the registry rule forbids inferring a ruling from a different application. Never-invent applies: this page draws the condition and names no part. PE to specify. ⚠ Note the geometry the PE inherits — see Figure 4: as drawn, only 2 in of the plate’s 4-in end actually faces the post, because of the registration question below.

⑥ ⭐ The two-inch question — what does a station mean?

The ring is the first thing this project has built that has to land on a gable line, and the two documents that place things there do not mean the same thing by “station 16”. The interiors register states wall extents — a wall “runs sta 16 → 43.5”, i.e. those numbers are faces. The as-built chain states member centerlines — geometry.py asserts station 16 is “to the cross CL”. They differ by exactly half a member: 2 in.

Why the page draws the wall-extent reading. It is the only one that keeps both ruled figures intact: R5’s eave plate stays one uncut 10'-2" stick, and register 0.4a’s ladder landing band stays 2'-2 1/4". The centerline reading grows the eave stick to 10'-6", pushes the ring 2 in past the roof’s own edge at both gable lines, and — the one that bites — leaves the west entry 1'-10 1/4" clear, below the owner’s 24-in NOMINAL ladder allowance, spending the 2¼ in of spare that 2.2b’s max-depth ruling was careful to preserve. ⚠ FIELD VERIFY applies to that allowance either way — it is NOMINAL until the ladder is bought.

⛔ OPEN — and it is one ruling, not four. The same 2 in shows up at every ring post:

the gable plate’s end gets 2 in of face against the ridge post instead of 4 — half the contact R21’s ends get;
corner posts 1, 2, 3 and 5 get 2 in of the doubled 2×8 nailer / 4×10 cross under them instead of 4;
post 4 gets 2 in of the N double trimmer (band 41.5→45.5, from TRIM_PLY_HALF_W);
and the west entry either keeps 2'-2 1/4" or drops to 1'-10 1/4".

⚠ The nailer’s own N–S registration is stated by no source at all — NAILER_D 8 is its depth; its 4-in width is derived from MEMBERS['nailer'] (“doubled 2×8 … rough true-2”). The offsets above are computed off the two members whose registration is in the datum: the cross and the ridge post. Owner ruling owed, then a Fable datum pass to write it down.

⑦ What it costs

The ring is two more sticks. Rough 4×4 × 12 ft at $12.80 each — 2 → 4 sticks, $25.60 → $51.20, a delta of $25.60. [MEMO 2026-08-14 §3 (R27, the price basis) · BOM: materials.xlsx tab Cabin (Year 2), the rough 4×4×12 row landed 2026-08-14.] Same Nelson rough line, same 2–3 week lead, no new SKU. ⚠ This counts plate stock only — the four new posts, the connectors and any gable framing above the plate are not priced here, because two of the three are not yet ruled.

⑧ Everything still open, in one list

⛔ The ring itself — PROPOSED. Owner ruling owed. If ruled it is a datum change (new members) → Fable v10 pass, then propagation (2.6a, the BOM, the set sheets, the cut list).
⛔ The registration — the two-inch question (Figure 4). One ruling settles four conditions. Owner, then Fable.
⛔ The side-framed plate-end-to-post connector. Unruled at R21’s two ends (raise 7) and at the four new gable-line ends. One detail, up to six joints. PE; nothing on file, nothing in the component registry.
⛔ The bearing under post 6 at station 69.75 — in a joist bay, over a tilted as-driven beam, 14.45 in north of a support that is itself NOMINAL and gated. PE.
⛔ The gable framing above the plate, on both lines. No source states it. It is also the load the north plate’s beam check needs.
⛔ The north plate’s member check (5'-2 1/2" and 2'-9 1/2" clear over the open porch face). Domain B; a new row, on the ruled gable framing.
⛔ The east porch segment (sta 16→43.5) still has no wall ruled — the frozen sibling study raised it and the ring does not answer it. What the ring does give that 27.5 in of east plate is a posted north end (post 1) where it had none; its south end is still R21’s unruled T-intersection, and no post is proposed there.
⚠ The corner post’s own cut length is unstated: 62 in from the floor if it runs through, or the 60.5 in stud pre-cut if it sits on the 1.5-in bottom plate. The frozen sibling study raised the same 60.5 figure against the workbook’s 61.5.
⚠ Two live figures for one eave run — 122 in (R24 / BOM) vs 121.25 in (2.6a ①). Raised in the sibling study, still open; it does not change any ring arithmetic (both buy 12 ft).

⑨ References — for the nonstandard parts of this ring

Prioritized as the house rule requires: code → definitive → trade.

2021 IRC R602.3.2 + Table R602.3.2 (single top plate, tied; the 3″×6″×0.036″ galvanised tie plate at corners and intersections) — enforced by the local building code for new permits. This is what makes a single-plate ring prescriptive at all, and it is the provision R27’s research established.
2021 IRC R104.11 (alternative materials and methods) — the route for R28’s half-lap, which is not a Table R602.3.2 tie.
Simpson Strong-Tie HTP37Z, 3″×7″ heavy tie plate — R28’s owner-specified ALTERNATE at the corners. ⚠ Only Canadian LSD factored resistances are on file; the US allowable row is not. Transcribed and bounded in the frozen sibling study; not repeated here, and no capacity is claimed on this page.
⛔ The side-framed plate end: NO SOURCE. Not in the code provisions above, not in any manufacturer literature on file, not in docs/components/. Never-invent — it is drawn as a condition and handed to the PE, unnamed.
⛔ The gable framing above the plate: NO SOURCE on this project. Drawn as an open triangle.

Deliberately NOT drawn

A recommendation · any adoption of the ring · a connector part, size, count or capacity for the side-framed joints · a pin, strap or fastener schedule at the corners · any member size, species or grade for the gable framing above the plate · a utilization figure for the north plate · a bearing detail under post 6 · a J-row · a component-registry ruling · a datum edit of any kind · a price for posts, connectors or gable framing · a resolution of the two-inch registration question · a correction to any source (a study reports; the owner rules; the owning document is fixed in its own commit).

Guards

build_study.py refuses to write the page if any of these fail, and names the offender:

the datum gate — geometry.py’s banner must still read DATUM v9, read off the banner itself;
the premise gate — PLATE_W = PLATE_DEPTH = POST_W = 4, PLATE_STEP = 0;
⭐ the ring must CLOSE: 66.5 + 4 + 37.5 = 108 exactly, and each gable line must fit one 12-ft stick;
⭐ the amendment’s premise — RIDGE_POST_FLOOR_TO_SOFFIT must equal Z_RIDGE_BOT and must clear the plate top, or the two-segment argument does not print;
the north plate’s east clear span must reproduce R21’s own 33.5 in, and the west entry must reproduce register 0.4a’s 2'-2 1/4" landing band;
⭐ the centered reading must FAIL the 24-in allowance — the whole two-inch finding turns on it, so if it ever passes the page must be rewritten, not re-run;
the registration offset must be exactly half a member, at all three conditions;
post 6’s footprint must fall between the two joist centerlines, or the bearing note says something false;
the four ridge reactions must still sum to the published total; the drawn soffit apex must sit below Z_APEX, which geometry.py forbids drawing as a soffit apex;
⭐ handedness — Figure 1 is a plan from above (east right, station down); Figure 2 looks EAST so south is on the right; Figure 3 looks SOUTH so west is on the right; and in every figure up is up;
every SVG label measured against its own viewBox, horizontally and vertically, on a deliberately generous character-width table;
a forbidden-claim scan (no adoption, no recommendation, no capacity, no connector selected) and a required-phrase scan (PROPOSED · FIELD VERIFY · PLANNING — NOT FOR CONSTRUCTION · the datum · R21 · R27 · R28 · R29 · R30 · R31 · never-invent · owner ruling owed · Domain B).

STUDY — NOT PART OF THE DESIGN. PLANNING — NOT FOR CONSTRUCTION. Frozen at DATUM v9 (2026-08-13), generated 2026-08-14 from treehouse/tools/draw/geometry.py by build_study.py, kept beside this page as provenance. ⛔ Do not re-run it on a later datum — write a new study instead. The ring is PROPOSED; the owner rules.

📁 archived · north-triangle-options — click to view

⭐ RULED — R35c (2026-08-15): station 16 goes full CATHEDRAL (deleted with the ring segments); R38 (2026-08-16): the 43.5 triangle — the one this study found on the real envelope, 0.043 sf from an enclosure threshold — is SHEATHED, with a closable screened vent high in it, removing the knife edge.
⛔ FROZEN at its own datum · 2026-08-15 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-15-north-triangle-options/

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V10 (2026-08-14) — ADOPTS NOTHING, SCHEDULES NOTHING, CLASSIFIES NOTHING — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. ALL THREE OPTIONS ARE PROPOSED — THE OWNER IS DECIDING, AND NOTHING ON THIS PAGE IS RULED.

The north gable triangle: open, screened, or sheathed

The triangle above the ring plate at station 16 — the roof’s own north edge, over the open porch face. Three options, the framing identical in all three, and one check that decides whether “open is free” or whether open re-prices the uplift chain. That check turns out to hinge on which of two north walls you are talking about, which is why this page begins by separating them.

⭐⭐ THE DISAMBIGUATION, FIRST — THERE ARE TWO NORTH WALLS AND THEY ARE 2'-3 1/2" APART.

	⭐ STATION 16 — THIS STUDY	⛔ STATION 43.5 — NOT THIS STUDY
what it is	the NORTH GABLE LINE — the roof’s north edge, over the OPEN PORCH FACE	R21’s door wall — the cabin’s actual north ENVELOPE
plate	the R32 ring’s north plate, two segments (5'-6 1/2" + 3'-1 1/2") side-framing the ridge post — J14, PE TO SPECIFY	R21’s single rough 4×4 plate, same Z 66, both ends the same unruled J14 condition
the member standing in it	N-GABLE ridge post — 127.9 lb, continuous floor → ridge soffit (8'-11 5/8")	N-MID ridge post — 806.1 lb, datum v10 rule D’s MERGED member (ridge post and door-wall west post, one stick)
above it	the triangle this page draws, and the 6" furring-framed rake (v10 rule C, NOMINAL — FIELD VERIFY at order)	a triangle of the same 20.25 sf that no source addresses at all — see the check below
below it	OPEN — the porch face, 5'-2" clear under the plate, ruled open by R32	the door, rough opening 32 × 55.75 in

Ridge reactions are the datum’s own [geometry.py datum v6 · CALC: calcs.xlsx 9_TH_RIDGE_3SPAN §5]; the four sum to 2165.5 lb, asserted in rederive().

⛔ PROVENANCE — a STUDY, and studies do not get refreshed.

class: STUDY (illustrative). Opposite staleness rule to the construction set — see studies/README.md and docs/decisions/2026-08-09-study-vs-set.md. Do not re-run this generator on a later datum; write a new study instead.
frozen on: DATUM v10 (2026-08-14) — read off geometry.py’s own banner at generation and asserted (DATUM_LIVE = v10).
generated: 2026-08-15
status: ⛔ ALL THREE OPTIONS PROPOSED. Not ruled, not adopted, not in the datum. The framing itself (P1) is proposed too. Nothing here writes back to any source.
workbook: calcs.xlsx RECALCULATED READ-ONLY 2026-08-15 (toolkit/build_v2 harness.recalc → LibreOffice headless; 0 error cells). The workbook was not edited and nothing computed here was written to it.
rulings behind it: R21 (the door-wall plate) · R27–R32 (the eave plates re-affirmed, the corner detail, C/H retired, solid corner posts, posts 4/6 located, the ring adopted) · R33/R34 (the four BOM calls; the ring registers to the as-built NAILERS, never the crosses or beams).
the proposals: docs/decisions/2026-08-15-gable-framing-and-connection-proposals.md — P1 (plumb studs + raked cap, both lines), P2 (this triangle; the memo recommends sheathed, with open/screened as the alternate), P10 (roof diaphragm boundary). Every one PROPOSED — PE to confirm; no R-number is issued by that memo.
adopts: Nothing. No option is selected, no member is sized, no capacity is claimed, no connector is named, no enclosure classification is computed, and no source is corrected.

① The two north lines, drawn

② The three options — the framing is identical, only the skin differs

P1 is the same in all three and the memo says so in terms: the open and screened variants retain the studs and the raked cap, because “the diaphragm boundary and the rake need the members regardless.” So the choice on this page is a skin choice, not a framing choice.

option	what fills the triangle	framing	what changes
(a) OPEN ⭐ owner preference	nothing — the porch bay is open to the roof cavity above the plate	P1 studs + raked cap — retained	the roof-cavity edge is exposed; the boundary member is unbraced by any panel
(b) SCREENED	insect screen between the studs, no structural panel	P1 studs + raked cap — retained	critters and birds excluded; structurally and for §26.2, identical to (a) — screen resists no pressure and closes no opening
(c) SHEATHED + SIDED the memo’s P2 pick	7/16 sheathing + siding to match the enclosure	P1 studs + raked cap — same	cavity closed; the gable-line boundary member gains a braced panel; the N plate picks up ~143.775 lb of triangle dead

③ ⭐⭐ THE DECIDING CHECK — ASCE 7-16 §26.2 enclosure classification

What the workbook actually carries today

No classification at all — deliberately. 2_LD_WIND row 182 states the convention in terms: “ENCLOSURE IS A PER-CONSUMER BINDING, NOT A TAB-LEVEL SETTING. No cell reads ‘the’ enclosure class — every consumer names which state it reads. There is deliberately NO single wind__gcpi name: the two states below are BOTH live, every design-pressure row below computes BOTH.” Both GC_pi values are live cells — ±0.18 enclosed and ±0.55 partially enclosed, ASCE 7-16 Table 26.13-1. Consumers that adopt one say so on their own row: the pile/uplift chain adopts ENCLOSED (9_TH_PILE_C F7, 2_LD_TAKEDOWN §5 — “that is the adopted classification… THE PARTIALLY-ENCLOSED ALTERNATES EXIST AND ARE NAMED”). The tab’s own note on GC_pi calls it “the biggest single demand lever” and hands the classification to PE/AHJ.

What the lever is worth, priced from the tab’s own two columns (no re-derivation — both columns are live and were recalculated today):

quantity	ENCLOSED (0.18)	PART-ENCLOSED (0.55)	Δ	cell
Per-post roof uplift, W only (strength)	468.3 lb	691.9 lb	+47.7%	C16 / C141
Net corner uplift, ASD 0.6D+0.6W	291.5 lb	425.6 lb	+46.0%	C99 / C100
C&C wall suction at the stud EWA, zone 4	-32.0 psf	-41.2 psf	+28.9%	C338 E/F
C&C wall suction, zone 5 (corner)	-39.5 psf	-48.7 psf	—	C339 E/F

Basis for all of it: V = 115 mph ult, Exposure C adopted, q_h = 25.00 psf [2_LD_WIND §V3.0 velocity-pressure ladder, C181]. The partially-enclosed column is not decoration — the whole uplift and hold-down chain would re-run on it.

The inventory the test needs, and what is in it

opening	wall	size	area	basis
SOUTH SE-3 egress window, 27 x 32 CLEAR	SOUTH sta 138	27 × 32 in	6.000 sf	RULED, R18 -- second means of egress of record
SOUTH gable pane G-A, fixed, west of the ridge post	SOUTH gable, above the plate	33.25 × 26.21 in	6.052 sf	RULED -- derived area-maximizing inscribed rectangle
NORTH door onto the covered porch (rough opening)	NORTH sta 43.5	32 × 55.75 in	12.389 sf	linked to 9_TH_HDR_DOOR; UPPER BOUND height
EAST window	EAST X 108	40 × 24 in	6.667 sf	RULED 2026-08-13 (R19) through the selector
§15 live sum, D565 / E565			4,479.54 in² = 31.108 sf	reconstructed here from the four rows and matched to the live cell — a build guard

⚠ ERRATUM FOUND WHILE READING §15 — RAISED, NOT FIXED (discrepancy protocol). The sum row’s own label reads “SUM — RULED openings entered above (SOUTH × 2 + NORTH). NOT the full set” and its note reads “Three openings, two walls. EAST is absent (under revision).” The live sum is four openings on three walls. It computes 4,479.54 in², which is 864 + 871.54 + 1,784 + 960 — the east 960 in² is in it. The east row did exactly what its own note promised (“The moment the width is ruled, this row and the sum below populate with no further edit”) when R19 ruled the width on 2026-08-13; the sum populated and the label describing it did not. Dated correction owed to Domain B. ⛔ Nothing was changed here.

⚠ AND THE INVENTORY IS INCOMPLETE — THE WORKBOOK SAYS SO ITSELF. 2_LD_WIND row 566, verbatim: “THE OPEN PORCH AND CHASE FACES — the covered porch’s north and west faces and the chase’s open faces are by far the largest openings on this structure, and their extents are PROSE-ONLY in geometry.py’s enclosure section, which deliberately declines to encode them… SO THE TWO CATEGORIES MOST LIKELY TO DECIDE THE CLASSIFICATION… ARE PRECISELY THE TWO THAT ARE NOT DIMENSIONED. Anyone computing 26.2 from this block alone would get a wrong answer confidently.”

Two further openings are missing that row 566 does not name, and both are consequences of rulings made after it was written: the west ladder entry R31’s jamb posts create (26.25 in clear × 62 in under the plate), and the triangle above the station-43.5 plate, which no source addresses in either direction.

⛔ SO EVERY AREA IN THE NEXT TABLE THAT IS NOT IN §15 IS DERIVED HERE, FOR THIS STUDY ONLY. It is computed from datum constants (WALL_SPAN, PLATE_Z, PLATE_DEPTH, POST_W, the station chain, WEST_ENTRY_CLEAR) purely so the §26.2 test can be shown to straddle its own threshold. Nothing below is entered in any source, written back to the workbook, or offered as a datum value — never-invent cuts both ways, and a derived area presented as an inventory figure would be exactly the confident wrong answer row 566 warns about. The derived areas: porch north aperture 41.33 sf net of three posts (46.50 gross) · porch east and west faces 11.84 sf each · west ladder entry 11.30 sf · either gable triangle 20.25 sf [the workbook’s own slope-defining triangle, 2_LD_WIND C92 / 2_LD_DEAD C50 — note it closes on Z_APEX 120, so the framed opening is smaller; the conservative direction for an opening-area test].

The test, worked on both live readings

ASCE 7-16 §26.2 makes a building partially enclosed only if all three hold: (1) A_o > 1.1 A_oi · (2) A_o > the lesser of 4 sf and 0.01 A_g · (3) A_oi/A_gi ≤ 0.20. Condition (2) clears by more than 15× on every row below (A_g = 49.5 + 20.25 = 69.75 sf → threshold 0.698 sf) and is guarded as a non-issue; condition (3) is bounded well under 0.20 by any plausible envelope area. Condition (1) — the dominant-opening test — is the only one that is ever close, and it is the one worked here, for wind on the north face, which is the governing direction for this question.

READING 1 — the envelope CLOSES at station 43.5 (the porch bay is outside it)

The covered porch is roofed but open on three faces (north, east and west); on this reading it is an open-sided extension and the cabin’s envelope is the door wall at station 43.5. ⭐ The station-16 triangle then appears in no term of the test — not A_o, not A_oi, not A_g, not A_gi. Options (a), (b) and (c) give identical arithmetic, and open is free.

row	condition assumed	glazing treatment	A_o	1.1 A_oi	result
1a	sta-43.5 triangle CLOSED	glazing counted	12.39 sf	33.02 sf	fails (1) → NOT partially enclosed
1b	sta-43.5 triangle CLOSED	glazing excluded (26.12.3)	12.39 sf	12.43 sf	fails (1) → NOT partially enclosed
1c	sta-43.5 triangle OPEN	glazing counted	32.64 sf	33.02 sf	fails (1) → NOT partially enclosed
1d	sta-43.5 triangle OPEN	glazing excluded (26.12.3)	32.64 sf	12.43 sf	meets (1) → PARTIALLY ENCLOSED

⚠ BUT READING 1 SURFACES A DIFFERENT TRIANGLE, AND THAT ONE IS ON A KNIFE EDGE Rows 1c/1d are the triangle above the station-43.5 plate — the cabin’s real north envelope, the same 20.25 sf because the roof is a constant N–S section. Closed, the north wall carries only its door (12.39 sf) and fails condition (1) either way. Open, it carries 32.64 sf — which still fails by 0.38 sf if the glazing is counted, and passes if the glazing is excluded under §26.12.3 (the site is not a wind-borne-debris region). The margin in row 1b is 0.043 sf on a 12.43 sf threshold. ⛔ No source states whether that triangle is open or filled. The proposals memo §1 details “both gable lines” — stations 16 and 138 — and is silent on 43.5.

READING 2 — the envelope INCLUDES the covered porch bay (one roof, one structure)

On this reading the windward wall for north wind is the porch’s open north face at station 16 — which is where this study’s triangle lives. The station-43.5 door becomes an interior opening and drops out of the envelope; the porch’s two side faces (11.84 sf each) become envelope openings and enter A_oi, which is what makes the threshold high enough to matter.

row	condition assumed	glazing treatment	A_o	1.1 A_oi	result
2a	sta-16 triangle SHEATHED (option c)	glazing counted	41.33 sf	59.07 sf	fails (1) → NOT partially enclosed
2b	sta-16 triangle OPEN / SCREENED (a, b)	glazing counted	61.58 sf	59.07 sf	meets (1) → PARTIALLY ENCLOSED
2c	sta-16 triangle SHEATHED (option c)	glazing excluded	41.33 sf	38.48 sf	meets (1) → PARTIALLY ENCLOSED
2d	sta-16 triangle OPEN / SCREENED (a, b)	glazing excluded	61.58 sf	38.48 sf	meets (1) → PARTIALLY ENCLOSED

⚠ AND THE FLIP IS ITSELF CONDITIONAL — read rows 2c/2d before quoting 2a/2b. The station-16 triangle decides the classification in exactly one of the four reading-2 cells: glazing counted (rows 2a/2b). If the glazing is excluded under §26.12.3 — which is the ordinary treatment outside a wind-borne-debris region — A_oi drops to 34.98 sf, the threshold falls to 38.48 sf, and the building classifies partially enclosed either way (rows 2c and 2d), because the 41.33 sf porch aperture clears it alone. So the honest statement is not “the triangle decides it” but: three inputs decide it, and the triangle is only the third.

⛔ Which envelope reading governs — reading 1 or reading 2. Nothing on file rules it.
⛔ How the glazing is treated — §26.12.3. Nothing on file rules it either, and it moves the threshold by 18.72 sf.
⛔ The station-16 triangle — decisive only under reading 2 with glazing counted, and only there.

⭐ That is a better answer than a verdict would have been: it tells the PE exactly which two rulings to make before the owner’s choice matters at all.

⭐⭐ THE ANSWER: “OPEN IS FREE” IS NOT ESTABLISHED — IT IS ONE OF TWO LIVE READINGS, AND THE OTHER ONE MAKES OPEN THE THING THAT RE-PRICES THE UPLIFT CHAIN Under reading 1 the station-16 triangle is outside the envelope and changes nothing — open is free, exactly as the owner’s preference assumes. Under reading 2 the same triangle straddles condition (1): sheathed, A_o = 41.33 sf against a 59.07 sf threshold and the building is not partially enclosed; open or screened, A_o = 61.58 sf and it is — a 4.3% overshoot, carried entirely by the 20.25 sf of triangle. That is GC_pi 0.18 → 0.55, i.e. +48% on per-post roof uplift and +46% on net corner uplift. ⛔ Which reading governs is the PE’s call and no source on file makes it.

⭐ A consequence worth stating plainly, because the memo does not make this argument. P2 recommends sheathing on boundary robustness, cavity closure and appearance. The stronger argument for (c) is the one above: sheathing removes 20.25 sf from the windward face’s opening on the one reading where the classification is live, and that classification is the workbook’s own “biggest single demand lever”. Symmetrically, (b) SCREENED buys nothing here: insect screen resists no pressure and closes no opening under §26.2, so (b) is (a) for enclosure purposes. ⛔ This is an observation about the arithmetic, not a selection — the owner rules.

④ The other four issues, worked

② Roof-diaphragm boundary — P10 lands in all three options

Confirmed from 2.6a §6.6, which pre-dates the memo and rules the members outright: “Blocking — Solid 2× between rafters, above the beam — and at the eave line, both planes,” placed there explicitly because “the 7/16 ZIP roof is element ③ of §6.5, and panel edges need a member to nail to at the boundaries.” That covers the ridge and eave boundaries. The gable-line boundary is P1’s raked cap — and the memo itself says the cap is retained in the open and screened variants, “the diaphragm boundary and the rake need the members regardless.”

⭐ So the boundary MEMBER and its nailing edge are present in all three options, and the memo’s “boundary robustness” argument for sheathing is weaker than it reads. ⚠ But it does not vanish, and this page will not overstate it: what sheathing adds is a braced panel behind the boundary member, and the memo’s own (b-2) carries the exact residual as a PE item — “the roof-diaphragm boundary then rides the raked cap + end-rafter blocking alone, without a sheathed panel bracing it: PE to confirm boundary adequacy in that configuration.” The argument narrows from “you need a boundary” to “you need the boundary braced”, and only the PE closes it.

⚠ Note the top boundary here is the sheathing plane, not an end-rafter soffit: 2.6a re-struck the rafter layout in 2026-08-06 so that no rafter falls in a gable-wall plane (owner constraint: both gable walls must die into the ceiling), insetting the end rafter from each gable line. The raked cap is therefore the only member on the gable line, in every option.

③ Weather — what comes through, and where it lands

Wind-driven rain and blowing snow enter high, through 20.25 sf of triangle at the north end, and fall into a covered porch that is 2'-3 1/2" deep (station 16 → 43.5). That is a narrow bay — the landing zone is the porch deck and the face of the R21 door wall, including the door itself. The triangle’s lowest point is at the eaves and its apex is at the ridge soffit (9'-10" above the floor), so anything driven through it has the whole porch depth to fall across and very little roof to fall under. ⛔ This is a lifestyle call, not a structural one, and the page does not weigh it — but it is the one issue on this list where the 2'-3 1/2" number does the deciding.

④ Critters and birds — why (b) exists

An open triangle against a ridge is the classic nesting geometry: sheltered, high, dry at the apex, invisible from the deck, and directly connected to the cold-roof cavity above the furring. Option (b) is exactly the answer to this and nothing else — it closes the aperture to wildlife while leaving it open to air, light and view. ⭐ Its cost is that it buys nothing on any of the other four issues: screen is not a diaphragm panel, not an opening closure under §26.2, and not a weather barrier.

⑤ What the workbook charges either way — and a second erratum

Wind sail: charged in full, regardless. 2_LD_WIND §1f derives the gable face from geometry as T1 (wall rectangle 49.5 sf) + T2 (triangle 20.25 sf) + the two rake/eave slivers = 71.434 sf, centroid 4.008 ft above deck, and that is “THE OVERTURNING FACE (§3 C20)”. The triangle is in it as solid area. So an open or screened triangle leaves the sail charge conservative — the model charges area that is not there. Nothing needs changing for (a) or (b); (c) simply makes the model true.

Triangle dead: ~143.775 lb, and today it is charged nowhere on the north. At dead2__wall_psf = 7.1 psf [2_LD_DEAD §6], 20.25 sf × 7.1 psf = 143.775 lb — the figure 2_LD_DEAD C78 already carries for the south triangle. On the north, C53 reads 0. Under (a) or (b) that zero stays right (there is no skin to weigh) and the charge is moot; under (c) a real figure is owed.

⚠ SECOND ERRATUM — RAISED, NOT FIXED. The memo points C53 at the wrong wall, and C53 itself has two identities.

The proposals memo §1.2 says adopting the sheathed north triangle “makes that row real”, citing 2_LD_DEAD C53. But C53 is the station-43.5 door wall, not the station-16 gable line — 2_LD_WIND row 394 pairs them explicitly: “NORTH wall (station 43.5) — door onto the covered porch … STRUCTURALLY FREE … No wall dead is charged for it either (2_LD_DEAD row 53).” Sheathing the station-16 triangle would need a NEW row; C53 would stay zero.
And C53 is described two ways in two cells that reference each other. Its own note calls it the “NORTH: door/porch gable”, while 2_LD_DEAD F50 treats row 53 as the north gable END, the counterpart of the SOUTH gable at station 138: “Applies at a gable END WITH a wall only (SOUTH, row 52); NORTH is structurally free (row 53).” Station 43.5 is not a gable end.
⭐ And station 16 — the actual north gable line — has no row anywhere. The six-line wall inventory (2_LD_WIND §11b) lists SOUTH 138, CHASE-SOUTH 96, EAST X108, CHASE-EAST X66.5, WEST X0 and NORTH 43.5. Before R32 there was no member on the station-16 line to inventory; there is now.

Three dated corrections owed — Domain B for the cells, Domain C for the memo’s cite. ⛔ Nothing was changed here.

⑤ ⭐ VERDICT PANEL — what would make each option defensible

(a) OPEN — the owner’s preference — is defensible as-preferred if, and only if, all three of these hold:

The PE reads the envelope as closing at station 43.5 (reading 1), so the station-16 triangle is outside it and cannot enter §26.2. This is the reading the workbook’s own wall inventory already implies — 2_LD_WIND row 369: “North of it is the covered porch, which the roof spans but no cabin wall encloses” — but it is implied, never ruled.
P10 lands, so the gable-line boundary member (the raked cap) exists regardless — which 2.6a §6.6 already rules for the ridge and eave lines and P1 supplies at the gable — and the PE confirms boundary adequacy with the cap unbraced by a panel, which is the memo’s own (b-2) residual.
The weather and wildlife consequences are accepted as lifestyle costs on a 2'-3 1/2" porch — or bought off with (b), which answers the wildlife half at no structural cost and no enclosure benefit.

⛔ What would FORCE (c) SHEATHED: the PE reading the covered porch bay as part of the envelope (reading 2). On that reading the open triangle is what carries A_o across 1.1 A_oi — 41.33 sf fails, 61.58 sf passes — and the building classifies partially enclosed, GC_pi 0.55. The whole uplift chain then re-runs at +48% per-post roof uplift, +46% net corner uplift and +29% on stud C&C suction — and the pile, hold-down and clip checks that currently bind the ENCLOSED column all move with it. On that reading (c) is not a preference, it is the cheap way out.

What would force neither: a PE ruling that the classification is partially enclosed anyway — because the porch face below Z 66 is 41.33 sf of permanent opening that R32 already ruled open, and no choice about the triangle can close it. If that is the answer, the triangle reverts to a weather-and-wildlife question and the owner’s preference stands on its own merits.

⛔ Open, unsourced or handed back — raised, not fixed

⛔ Nothing was corrected in any source from here. A study reports; the owner rules; the owning document is fixed in its own commit (discrepancy protocol).

⛔⛔ NEEDS_CONTEXT — which reading of the envelope governs. No source on file rules whether the covered porch bay is inside or outside the building envelope for §26.2. It decides this study’s whole question and it is a PE/AHJ call, exactly as 2_LD_WIND F8 says.
⛔⛔ NEEDS_CONTEXT — is the triangle above the station-43.5 plate open or filled? No source addresses it in either direction. It is 20.25 sf on the cabin’s real north envelope — bigger than the door — and rows 1b/1c/1d above show it sitting on the §26.2 threshold. Owner/PE. ⛔ Never-invent: the page draws no answer for it.
⛔ The §15 inventory cannot decide §26.2 as it stands — the workbook says so (row 566) and two further openings created by later rulings (the R31 west ladder entry; the station-43.5 triangle) are absent too. Domain B, when the enclosure read lands.
⚠ §15 sum-label erratum (four openings, three walls, not three/two) — Domain B, dated correction.
⚠ C53 mis-cite and double identity (the memo’s §1.2 cite; the C53/F50 gable-vs-door-wall conflation; station 16 has no row at all) — Domain B + Domain C.
⛔ J14 — the side-framed plate-end-to-post connector is named by no source. Six joints, one application, PE TO SPECIFY (geometry.CONNECTIONS). It is the joint at both ends of both segments of this triangle’s own plate. Unchanged by anything here.
⚠ The rake is NOMINAL — 6" at both gables, v10 rule C, with [PE/MFR] max panel rake overhang + rake trim detail FIELD VERIFY AT ORDER still open.
⚠ The out-of-plane stud check is a basis, not a check. The memo’s §1.1 states the C&C route and a worst stud reaction of ~66 lb enclosed / ~81 lb part-enclosed; the Domain B row is not written, and it would be written on whichever enclosure state the PE rules.
⚠ R34 registration is FIELD-PENDING. The ring registers to the as-built NAILERS, never the crosses or beams; the N nailer strays ~1.22 in north at the west end (NAILER_N_STRAY_W_DERIVED, from N_CROSS_FAN_DEG 0.644°). FIELD VERIFY — tape the nailer faces. Nothing on this page is blocked on it.

Deliberately NOT drawn

A recommendation · any selection among (a), (b) and (c) · an enclosure classification · any correction to calcs.xlsx, geometry.py, the proposals memo or any register · a connector, part number, size, count or capacity for J14 or for anything else · a member size, species or grade beyond what P1 already proposes · a utilization figure for the north plate (that row is Domain B, still owed) · a J-row · a component-registry ruling · a datum edit of any kind · a price · a ruling on the station-43.5 triangle · any modification of the two frozen sibling studies, which are quoted only.

Guards

build_study.py refuses to write the page if any of these fail, and names the offender:

⭐ the datum gate — geometry.py’s banner must still read DATUM v10, read off the banner itself
⭐ the premise gate — PLATE_W = PLATE_DEPTH = POST_W = 4.0, PLATE_STEP = 0
⭐ the disambiguation’s own premise — the two north lines must stay 2'-3 1/2" apart and the two ridge posts must carry different loads, or the page’s central claim needs re-founding
the gable line must close: 66.5 + 4 + 37.5 = 108 exactly, and the east clear span must still reproduce R21’s own 33.5 in
the ridge post must be continuous past the plate at all four stations, or P1’s studs-beside-not-bearing-on premise is false
⛔ the drawn soffit apex must sit below Z_APEX, which geometry.py forbids drawing as a soffit apex — the trap the R20-held door memo fell into
the tallest gable stud must still derive P1’s stated ~50 in, and the west stud must be the taller of the two
⭐ the §15 sum reconstructed from the four rows must match the workbook’s own live cell to 10^-6 in², or the erratum claim is unfounded and must be re-checked before it is published
⭐⭐ the finding must still straddle — reading 2 must FAIL condition (1) sheathed and MEET it open, and reading 1 must straddle on the station-43.5 triangle. If either stops straddling the page must be rewritten, not re-run
condition (2) must clear by > 15× on every row, or the page’s “only condition (1) is close” wording is wrong
⭐ handedness — Figure 1 looks EAST so south is on the right; Figure 2 looks SOUTH so west is on the right; Figure 3 is a plan from above with east right and north up; and up is up in both elevations
every SVG label measured against its own viewBox, horizontally and vertically, on a deliberately generous character-width table
a forbidden-claim scan (no adoption, no selection, no capacity, no classification computed, no connector named) and a required-phrase scan (PROPOSED · FIELD VERIFY · PLANNING — NOT FOR CONSTRUCTION · the datum · the freeze rule · DERIVED HERE · never-invent · NEEDS_CONTEXT · Domain B · R21 · R32 · R34 · P1 · P2 · P10 · J14 · ASCE 7-16 §26.2)

STUDY — illustrative, frozen at DATUM v10 (2026-08-14), generated 2026-08-15 from geometry.py and a READ-ONLY recalculation of calcs.xlsx. PLANNING — NOT FOR CONSTRUCTION. All three options PROPOSED; the owner rules. ⛔ Do not re-run this generator on a later datum — write a new study.

📁 archived · post-bases — click to view

⭐ RULED R35 (2026-08-15, all pending PE) — AC4RZ cap pairs (MAX nailing) at the mid stations, EPC4RZ at the gables (both field-sprayed black), 2 × MSTA12Z strap pairs stand at the bases (caps-as-bases rejected: AC not listed as a base; BC40R fails coating/capacity), and the nailer-to-cross and trimmer-to-cross shim joints get the same clustered MSTA12Z tie.
⛔ FROZEN at its own datum · 2026-08-15 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-15-post-bases/

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V10 (2026-08-14) — ADOPTS NOTHING, SCHEDULES NOTHING — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. EVERY DETAIL HERE IS PROPOSED (P5 / P6 / P7) — J12 AND J15 ARE BOTH PE TO SPECIFY.

The four post-base conditions

Ten posts stand on this deck — four ridge posts and six ring posts — and they do not land the same way. This page draws all four base conditions side by side, on the datum, with the demand and the seat arithmetic each one actually carries.

⭐⭐ THE OWNER’S QUESTION, CONFIRMED — AND SHARPENED.

Yes: the six ring posts and the four ridge posts have different base conditions. But the dividing line is not “ridge vs ring”, and the four ridge posts are not alike either. The real distinction is whether the member is CONTINUOUS through the floor plane or STARTS on it:

The four ridge posts are continuous. RIDGE_POST_FLOOR_TO_SOFFIT = 8'-11 5/8" holds at all four stations (2.6a, datum v6 rule D) — each is one stick from its bearing chain to the ridge soffit. A continuous member cannot stand on a bottom plate, so there is no bottom plate anywhere in conditions ① and ②.
The six ring posts start on the floor. P7 proposes they land ON the 1½-in PT bottom plate, which is exactly why the ruled cut is 5'-0 1/2" — plate underside 62 less the bottom plate 1.5 (R33-A).
And the ridge posts split again, because what is under the floor at a gable station (a doubled 2×8 nailer on the Year-1 cross) is not what is under it at a mid station (a double trimmer that must be bypassed). Different chain, ~7× the demand.

Four conditions, then — and a fifth would exist if post 6 were counted separately from the other five ring posts, which is how this page treats it, because its bearing is the one that is FIELD-PENDING.

⛔ PROVENANCE — a STUDY, and studies do not get refreshed.

class: STUDY (illustrative). Opposite staleness rule to the construction set — see studies/README.md and docs/decisions/2026-08-09-study-vs-set.md. Do not re-run this generator on a later datum; write a new study instead.
frozen on: DATUM v10 (2026-08-14) — read off geometry.py’s own banner at generation and asserted (DATUM_LIVE = v10).
generated: 2026-08-15
status: ⛔ EVERY DETAIL PROPOSED. P5 (gable seats), P6 (mid seats), P7 (ring post bases) of the 2026-08-15 proposals memo, which issues no R-numbers. geometry.CONNECTIONS carries J12 (ridge post base) and J15 (ring post base) both as PE TO SPECIFY, and the memo does not change that.
workbook: calcs.xlsx RECALCULATED READ-ONLY 2026-08-15 (toolkit/build_v2 harness.recalc → LibreOffice headless; 0 error cells). The workbook was not edited.
rulings behind it: R30 (all ring posts solid rough 4×4) · R31 (posts 4 and 6 located — they jamb the west entry) · R32 (the ring adopted) · R33-A (the 5'-0 1/2" cut) · R34 (⭐ the ring registers to the as-built NAILERS, never the crosses or beams).
adopts: Nothing. No seat is specified, no connector part is named, no capacity is claimed beyond the workbook figures it cites, and no source is corrected.

① Which post is which condition

② The four conditions, side by side

⛔ These are STACK DIAGRAMS, not dimensioned sections. Member depths are the datum’s own (NAILER_D 8, CROSS_D 10, JOIST_D 8, DROP_BEAM_D 10). The below-floor seat elevations are stated by no source — that is precisely what J12 asks the PE for — so none is drawn, and a build guard refuses the page if any Z value appears below the floor line.

③ Condition ① — RIDGE POST at a GABLE station (16 / 138)

Chain: the continuous ridge post bears directly on the doubled 2×8 gable nailer → a scribed shim pack → the Year-1 flush 4×10 cross. ⛔ No bottom plate.

Why a shim at all — it is already required, before this proposal. The nailer’s own arithmetic demands it: 0.985 utilization gap-open on the no-cull basis, falling to 0.333 shimmed [LIVING memo §13]. P5 only details a shim the structure already needed.
What P5 proposes: a shim pack of exterior-grade structural plywood, plies HORIZONTAL (load through the panel thickness), full nailer-width × full 4-in post footprint minimum, field-scribed to the computed crown gap; galvanized steel plate as the equal alternate at thin gaps. The rule stated so it cannot be built wrong: no end-grain shim, no tapered softwood-shingle pairs under a structural seat.
⛔ The shim MATERIAL is open, and the memo says so. §7: “no code provision and no manufacturer/engineering literature found that names a shim material for a wood post-bearing seat” — nearest kin is FEMA P-804’s “compression blocks”. The geometry is proposed firmly; the material is tier-3 trade practice and the PE rules. This page draws no thickness for it, and a guard enforces that.
The tie: two flat vertical straps per post, post face → nailer face, fasteners clustered within ~3–4 in of the joint line (the 2.6a §6.4 green-shrinkage rule). ⛔ No part is named on the drawing — the memo proposes an MSTA12Z-class strap, and the component registry forbids inferring a ruling across applications: MSTA12Z’s only registry application is J4, which is gated on the now-RETIRED C/H scheme. A NEW dated application row is owed, PE TO CONFIRM.

④ Condition ② — RIDGE POST at a MID station (43.5 / 96)

Chain: the continuous ridge post → full-depth squash blocking flanking the double trimmer on BOTH sides → direct to the NEW 4×10 cross. ⛔ No bottom plate, same reason as ①.

⭐ The trimmer is BYPASSED, not loaded. The load may not enter the trimmer in bending [LIVING §11] — the blocking takes it straight past to the cross. The two-sided fit also avoids the one-sided prying mode the guard-base history documents.
Reused, not invented: this is the same detail already installed under the ridge posts (set 7d / Cabin r82).
Both mid posts are MERGED members and their heads are PACKED TIGHT (owner ruling 2026-08-08 — 2.12 §11d reversed): S mid is the chase column (datum v7 rule A, one member two jobs); N mid is the door-wall west post (datum v10 rule D). So the base seat is where all vertical adjustment lives — there is no head gap to take it up. That is what makes condition ② the demanding one.
⚠ What rides on this seat: the §10.3 differential-shrinkage stack turns on the per-station cross-grain height, which turns on this detail — a named PE question carried on J12 since datum v6. This proposal puts a concrete detail under it; it does not answer it.

The demands, and the seat arithmetic

ridge post	station	condition	demand (D+S)	seat utilization	verdict
N GABLE	sta 16	①	127.9 lb	0.020	PASS — 51×
N MID	sta 43.5	②	806.1 lb	0.124	PASS — 8×
S MID	sta 96	②	959.9 lb	0.148	PASS — 7×
S GABLE	sta 138	①	271.6 lb	0.042	PASS — 24×
SUM	asserted in `rederive()` against `w × RIDGE_SPAN_CC`		2165.5 lb	the four stations close on the published total

Demands [MEMO 2026-08-15 proposals §2 · CALC: calcs.xlsx 9_TH_RIDGE_3SPAN §5]. Seat arithmetic [CALC: '9_TH_POST_RIDGE' §BEARING PERP (rows 43–55)]: bearing area 16 in² (the full 4×4 footprint), F_c⊥′ green = 555 × C_M 0.67 × C_b 1.09375 = 406.71 psi (NDS Supp. Table 4A wet service, no exception taken) → allowable 6,507 lb per seat. Worst station 0.148. Both seats of every post (cross below, ridge soffit above) are numerically identical — same area, species, derate and load. Slenderness l_e/d = 16.5 ≤ 50 (NDS 3.7.1.4, PASS); governing axial utilization 0.0935, so bearing governs the member, not buckling.

⑤ Condition ③ — RING / JAMB POST (six posts)

Chain: a 5'-0 1/2" solid rough 4×4 standing ON the 1½-in PT bottom plate, on the subfloor, over full-depth squash blocking in the bay below, down to its own bearing chain.

⭐ The bottom plate is why the cut is 5'-0 1/2" — plate underside 62 less the 1½-in bottom plate (R33-A, the same figure as the ruled stud pre-cut). Under the alternate (plate-interrupted) the post would be 62 instead. P7 proposes post-on-plate, and keeps the plate continuous past the post: one wall-assembly detail everywhere, continuous sill seal, continuous shear-wall bottom-plate nailing, no notched plate ends to detail.
Why the bearing is not a problem: P7 states it at a bounding 1,500 lb corner reaction — f_c⊥ ≈ 94 psi against a deliberately pessimistic wet floor of 372 psi (555 × C_M 0.67, the green Eastern Hemlock–Tamarack basis; any PT southern pine plate is denser) → ≈ 0.25 utilization. ⚠ That is the memo’s bounding figure, not a computed reaction — no ring post has a computed demand anywhere (see the open items).
⭐ R34 — registration. The corner posts center on the as-built NAILER lines, never the N–S crosses or the E–W beams, which are not square. The owner, from the field: “in order for cabin to square the posts need to be centered on double nailers.” The datum corroborates: the N cross fans 0.644° off the rim square, so a square line on it is flush at the east end and 1.22 in proud at the west (NAILER_N_STRAY_W_DERIVED). Tape precision remains FIELD-PENDING — FIELD VERIFY — but nothing is blocked on it.
Post 4 is the exception among the five: it lands on the N double trimmer at the 1.2e wall’s south end (R31), not on a nailer.
The tie: two flat vertical straps per post, one per available outer face, post face → rim/nailer/trimmer face, clustered at the joint line. At the four plate corners these sit below the R28 belt-and-suspenders corner straps (post top → plate) — same family, same face, one detail language top and bottom.
⭐ The absence of a formed post base was RESEARCHED, not assumed (§3.2): Simpson’s formed caps/bases are U-saddles made to surfaced 3½-in widths and a true-4.0 rough post will not seat; the FPBB44 is a fence-retrofit part (wrong finish, wrong substrate); AWC DCA 6 publishes no flat-strap post attachment and its post-to-beam family does not fit a flat true-4 base on a deck. So the strap pair rests on the tension-strap family precedent — legitimate precisely because gravity stays wood-to-wood bearing and the straps carry tension only. The PE confirms an assembly, not a SKU.

The six ring posts

post	role	location	bears on	condition
POST 1	NE corner	sta 16 · X 104	N nailer + cross chain	③
POST 2	SE corner	sta 138 · X 104	S nailer + cross chain	③
POST 3	SW corner	sta 138 · X 0	S nailer + cross chain	③
POST 4	W entry N jamb	sta 43.5 · X 0	N DOUBLE TRIMMER (1.2e wall S end)	③
POST 5	NW corner	sta 16 · X 0	N nailer + cross chain	③
POST 6	W entry S jamb	sta 69.75 · X 0	J5/J6 joist BAY → W drop beam	④

Posts 4 and 6 jamb the west ladder entry: 43.5 → 69.75 = 26.25 in clear, at or above the 24-in ladder allowance (R31/R32, asserted). They also sit exactly at the ends of the ruled unsupported west-eave run — the ring lengthens no eave span; it gives two already-assumed supports a named post.

⑥ Condition ④ — POST 6, the special case (station 69.75)

Everything in condition ③, except what is under the blocking.

⛔ POST 6 IS THE ONE BASE WITH A REAL GAP, NOT JUST AN OPEN TAG.

Its footprint is 69.75 → 73.75, which falls in the bay between joists J5 (64.5) and J6 (75) — no joist lands under it. That is the entire reason it needs its own condition, and it is asserted in the build: the drawn footprint and the computed footprint must both fall inside the bay or the page refuses to build.
P7 proposes full-depth squash blocking in that bay, post → block → the west DROP beam line — the ridge posts’ own installed detail (set 7d / Cabin r82) reused, not invented.
⚠ The as-driven west beam position under the footprint is FIELD-PENDING (v10 rule G). The beam is tilted as driven.
⚠ And the beam’s own nearest support going south is itself NOMINAL: the planned Year-2 west-mid TAB, 14.45 in away, root-blocked at probe, PE + arborist gated. ⭐ Frame warning, guarded here: W_YR2_MID_STATION_PLANNED_NOMINAL = 53.8 is measured up-string off the SW post (tab §6 B38 / 1.6 §3), not from the north rim like every NS_STATION — it converts to station 84.2 on the N-rim chain. Reading it as 53.8 would put the TAB 15.95 in on the wrong side of post 6. The conversion is asserted against the frozen sibling ring study’s own figure.

⭐ THE COMMON PANEL — what is true of all four

Gravity is by BEARING, everywhere. Wood to wood, through full footprints. The project’s all-bearing rule is not a preference here; it is what lets flat straps be legitimate on rough true-4 stock at all.
Hardware resists SEPARATION and UPLIFT only. No strap, screw or plate on this page carries a pound of gravity. A screw-only base (an SDWS toe pair) is not proposed as the tie — an inclined screw resists uplift in withdrawal-dominated action, the exact class the F7 purge removed.
Rough true-4 stock takes FLAT hardware only — no formed saddles. That is the J3/HUCQ lesson, and it is why every tie here is a flat strap.
Green hemlock ⇒ ZMAX/G185 straps + HDG/SD fasteners, and fasteners clustered at the joint line (2.6a §6.4 green-shrinkage detailing).
⛔ All four details are PROPOSED. The PE confirms. geometry.CONNECTIONS carries J12 (ridge post base → bearing chain) and J15 (ring post base → deck bearing chain) both as PE TO SPECIFY, and it says in terms that the seat detail below the floor “is stated by NO source”. Nothing in the proposals memo changes those statuses, and nothing on this page does either.
⭐ J14 is marked where it belongs — and it is NOT a base. Where a ring plate meets a continuous post, nothing bears on the post’s top (v6 rule D), so the plate side-frames into the post face: that is J14, six joints, one application, PE TO SPECIFY, named by no source. It occurs at the tops of the four ridge posts, never at any base on this page. Do not confuse the two asks.

⛔ Open, unsourced or handed back — raised, not fixed

⛔ Nothing was corrected in any source from here. A study reports; the owner rules; the owning document is fixed in its own commit (discrepancy protocol).

⛔ J12 and J15 remain PE TO SPECIFY. P5/P6/P7 give the PE a concrete detail to confirm instead of an open question — that is the whole point of the proposals memo — but they are proposals.
⛔ The gable-seat shim MATERIAL (plywood pack vs steel plate) is open after exhausting the source ladder. PE rules. No thickness is drawn here.
⛔ No listed flat post-BASE part exists for true-4 rough posts — researched and reported in §3.2, not assumed. The PE confirms an assembly, not a SKU.
⛔ NEW component-registry application rows are owed before any strap is real: J12-gable (ridge post base → gable nailer), the J12 mid equivalent, and J15 (ring post base → deck). MSTA12Z’s only existing application is J4, gated on the RETIRED C/H scheme, and a ruling there does not transfer.
⚠ No ring post has a computed demand. P7’s bearing check runs on a bounding 1,500 lb corner reaction; the ring posts’ actual reactions are not computed anywhere. Domain B row owed — and it would come with the north-plate and south-collector rows the ring already owes.
⚠ Post 6’s bearing is FIELD-PENDING twice over — the as-driven beam line under the footprint, and the NOMINAL, gated TAB 14.45 in south of it. FIELD VERIFY.
⚠ R34 tape items: the nailer faces vs the cross centerlines at both ends, and the W-beam line at station 69.75. Documentation, not blockers — the model no longer pretends the crosses are the reference. FIELD VERIFY.
⚠ The §10.3 differential-shrinkage stack rides on condition ②’s seat and is a named PE question on J12 since datum v6. Unchanged here.

Deliberately NOT drawn

A recommendation · any adoption of P5, P6 or P7 · a connector part number, size, count or capacity · a shim thickness · any below-floor seat elevation · a fastener schedule · a component-registry ruling · a J-row · a datum edit of any kind · a price · a computed ring-post reaction · a correction to any source · any modification of the frozen sibling studies, which are quoted only.

Guards

build_study.py refuses to write the page if any of these fail, and names the offender:

⭐ the datum gate — geometry.py’s banner must still read DATUM v10, read off the banner itself
the premise gate — every post is a solid rough true 4×4, because the whole seat arithmetic is computed on its 16 in² footprint
⭐ the ridge posts must be CONTINUOUS past the plate, or conditions ① and ② collapse into ③ and the page’s central claim is false
⭐ conditions ①/② must carry NO bottom plate and ③/④ must carry one — the owner’s actual question, so it may not silently invert
the ring cut must still derive the ruled 5'-0 1/2" from plate underside less bottom plate
the four ridge reactions must still sum to the published 2165.5 lb, and the worst seat utilization derived here must reproduce 9_TH_POST_RIDGE C50 to nine decimals — if the page cannot reproduce a figure, it may not print it
F_c⊥′ must still be 555 × C_M × C_b on the tab’s own factors
⭐⭐ post 6 must fall in the J5/J6 bay — both computed and as drawn, checked separately, because a drawing that agrees with its own prose by accident is not a check
⭐ the TAB frame conversion — W_YR2_MID_STATION_PLANNED_NOMINAL is in the SW-post frame and must convert to station 84.2 on the N-rim chain, matched against the frozen sibling study’s figure
post 4 must still coincide with the N double trimmer; the west entry must still clear the ladder allowance
both merged members must still be on the ridge line (v7 rule A / v10 rule D)
⛔ the never-invent guard — no Z value, elevation string or shim thickness may be drawn below the floor plane in Figure 2, because no source states one and J12 is what asks for them
⭐ handedness — Figure 1 is a plan from above with east right and north up
every SVG label measured against its own viewBox, horizontally and vertically, on a deliberately generous character-width table
a forbidden-claim scan (no adoption, no part selected, no capacity asserted, no seat specified) and a required-phrase scan (PROPOSED · FIELD VERIFY · PLANNING — NOT FOR CONSTRUCTION · the datum · the freeze rule · never-invent · PE TO SPECIFY · Domain B · J12 · J14 · J15 · R30 · R31 · R33 · R34 · P5 · P6 · P7)

STUDY — illustrative, frozen at DATUM v10 (2026-08-14), generated 2026-08-15 from geometry.py and a READ-ONLY recalculation of calcs.xlsx. PLANNING — NOT FOR CONSTRUCTION. Every base condition PROPOSED; J12 and J15 are PE TO SPECIFY. ⛔ Do not re-run this generator on a later datum — write a new study.

📁 archived · cathedral-porch — click to view

⭐ RULED R35c (2026-08-15) — the FULL CATHEDRAL PORCH ADOPTED pending PE: the two station-16 ring-plate segments deleted, the porch open deck-to-rafters (ridge-line headroom 5'-2" → 8'-11⅝"). Executed as datum v11; the raised consequences (N roof-edge diaphragm boundary → PE; the N-gable post's doubled unbraced height → Domain B) ride with it as scheduled items.
⛔ FROZEN at its own datum · 2026-08-15 · kept whole here; repo: treehouse/design-and-reference/studies/2026-08-15-cathedral-porch/

STUDY — NOT PART OF THE DESIGN — FROZEN AT DATUM V10 (2026-08-14) — ADOPTS NOTHING, SCHEDULES NOTHING, SELECTS NO CONNECTOR — GENERATED FROM geometry.py, NOT HAND-DRAWN

PLANNING — NOT FOR CONSTRUCTION. NOT A STAMPED DESIGN. THE PROPOSAL ON THIS PAGE IS PROPOSED — THE OWNER RULES, AND THE DATUM IS v10 AND STAYS v10 UNTIL A FABLE PASS WRITES v11.

The full cathedral porch

The owner’s proposal: the porch region — stations 16 to 43.5, north of the door wall — fully open and unobstructed from the deck floor to the rafters. Concretely, the two north gable-line ring-plate segments are deleted, so nothing crosses the porch’s north face at Z 66. He saw the 5'-6″ plate over the north nailer in the frozen triangle study and rejected the headroom. This page draws it, and works the consequences honestly — including two the proposal did not name.

⛔ WHAT IS PROPOSED, AND WHAT IT AMENDS. If adopted this amends datum v10 rule A — the R32 ring — and needs a v11 Fable pass. It is labelled that way throughout. Deleted: the two station-16 gable plate segments (GABLE_PLATE_SEG_W 5'-6 1/2" + GABLE_PLATE_SEG_E 3'-1 1/2"), and with them the gable triangle above and everything that stood in it. Kept at the north face: the N-GABLE ridge post (continuous, 127.9 lb); the NW and NE corner posts (R30 solid rough 4×4, now carrying the eave plate ends); and, per the owner, “associated supporting NW/NE 16-oc framing up to the rafters” — read here as a narrow framed strip at each corner, the middle of the face fully open. ⛔ No source states the strip width. It is drawn ILLUSTRATIVE and tagged OPEN; the owner rules it.

⭐ The headroom it buys, at the ridge line: 5'-2" → 8'-11 5/8" — a gain of 3'-9 5/8". And a band 7'-8" wide where a six-foot adult clears the ceiling standing up. Under the deleted plate that band is zero: the plate’s underside is 5'-2" across the entire face.

⛔ PROVENANCE — a STUDY, and studies do not get refreshed.

class: STUDY (illustrative). Opposite staleness rule to the construction set — see studies/README.md and docs/decisions/2026-08-09-study-vs-set.md. Do not re-run this generator on a later datum; write a new study instead.
frozen on: DATUM v10 (2026-08-14) — read off geometry.py’s own banner at generation and asserted (DATUM_LIVE = v10).
generated: 2026-08-15
status: ⛔ PROPOSED. Not ruled, not adopted, not in the datum. The framing it would replace (P1 / P2) was itself only proposed. Nothing here writes back to any source.
workbook: calcs.xlsx read READ-ONLY. ⚠ The repo file carries no cached formula values (it was last written by openpyxl and has no xl/externalLinks/ part), so formulas and note text were read from the repo file and live values from a headless LibreOffice recalculation of a scratchpad copy (0 error cells). The repo workbook was not opened for writing and nothing here is written back.
rulings behind it: R5 / R27 (the eave plates) · R28 (the corner detail) · R30 / R31 (solid corner posts; posts 4 and 6) · R32 (the ring adopted) · R33 / R34 (the BOM calls; the ring registers to the as-built nailers) · R21 (the door wall, untouched) · R10 / R7 / R8 (the counted shear lines and the porch west wall).
the proposals: docs/decisions/2026-08-15-gable-framing-and-connection-proposals.md — P1 (plumb gable studs + raked cap), P2 (the north triangle), P4 (the N/S plate load basis), P9 (diaphragm chords), P10 (roof diaphragm boundary). Every one PROPOSED; that memo issues no R-number, and neither does this page.
siblings: Three FROZEN studies are quoted only and not modified: 2026-08-14-plate-ring-layout, 2026-08-14-plate-corner-connections, 2026-08-15-north-triangle-options.
adopts: Nothing. No option is selected, no member is sized, no capacity is claimed, no connector is named as chosen, no enclosure classification is computed, and no source is corrected.

① The north gable line, before and after

② The cathedral volume — section looking north

③ The ring becomes a U

④ What the north plate’s jobs were, and where they go

The plate had seven jobs on file. One transfers to a joint nobody has specified, one turns out never to have existed as work, and the other five come out clean.

the job	what it was under the v10 ring	where it goes under the proposal
(a) ring closure & the corner tie at NE / NW	The two north plate segments met the eave plates over the R30 corner posts in the R28 half-lap — scheduled J13.	⭐ Becomes an EAVE-PLATE-END-to-CORNER-POST condition. The perpendicular member disappears, so there is nothing to lap and nothing to tie to in the E–W direction: the eave plate simply runs over the post. J13 goes from four gable-line corners to two (south only). ⛔ The connection that replaces it is named by no source.
(b) the R28 half-lap itself, at the north corners	Dissolves. R28's lap handedness was derived, not chosen (the rafter seat sits inside the lap square, so the eave plate takes the upper tongue) — that derivation needed two plates.	One plate, no lap. ⭐ The south corners keep theirs unchanged, so R28 is not reversed — it is halved in scope. The scoped owner exception to the no-site-built-joinery position therefore covers two laps, not four.
(c) gable-triangle stud bearing	P1's plumb 2×4 gable studs stood on the plate's 4.0-in top face; P2 was the choice of what skinned them.	Moot — there is no triangle. The face is open. The frozen sibling study's three options (open / screened / sheathed) all presupposed the plate and the studs; this proposal deletes the question rather than answering it.
(d) lateral — is any Σb lost?	No. The station-16 line was never a counted shear line, and it is stronger than that: it is not in the wall inventory at all.	`2_LD_WIND` §11b “THE SIX WALL LINES, AND WHICH MAY BE COUNTED”, column COUNTED IN Sigma-b? — the counted E–W lines are SOUTH (station 138) and CHASE-SOUTH (station 96, R10); the counted N–S lines are EAST and WEST (R7). The northernmost line the inventory knows is the cabin NORTH wall at station 43.5, and that row is length zero, uncounted, “STRUCTURALLY FREE”. ∴ no Σb term changes.
(e) collector / chord duty on the north edge	The E and W eave plates run continuous over the porch regardless — the rafters bear on them — so the N–S chords/collectors are unaffected.	And the N–S chords are not at the plate line at all: `2_LD_WIND` §12h item (4) says “the rim and the doubled gable nailers are the chords by default” — deck-level members. Deleting a member at Z 66 touches neither. ⚠ What it does not fix is the standing gap: §14.7 — “in the porch region those edges have no continuous chord member identified… UNCHECKED and UNNAMED.”
(f) the “double duty as door header” job	R32's own words for the north plate. The proposals memo §1.4 gives the west segment no floor or roof bearing, “so the triangle IS its load.”	⭐ No triangle, no header, nothing to span — and therefore nothing to check. The v10 rule G Domain B row (“the N plate's beam duty over the open porch face, clear spans 62.5 / 33.5”) is withdrawn before it was ever written: a workbook-wide search finds zero occurrences of 33.5 and no north-plate row in `9_TH_SCHED`’s 40 instances.
(g) J14 — the side-framed plate-end joint	Six joints, one application, PE TO SPECIFY: R21's door-wall plate at both ends, plus both gable lines’ two segment ends into the continuous ridge posts.	⭐ Six becomes four. The two north-segment ends go with the segments. The unruled PE ask shrinks; it does not go away, and it is replaced by a different unruled joint at the two north corners (row (a)).

⑤ What does not change — and the owner should see this list before the other one

⭐ The porch does not become a different building. Both eave plates still cross it at Z 66, because they are what the rafters bear on. Everything below is verified against the live sources, not assumed.


The WEST eave plate, crossing the porch at Z 66	THE rafter bearing, and untouched by this proposal (R5, re-affirmed R27). It sits on register 1.2e’s framed bearing wall from station 16 to 43.5 — the wall with the two open ports. ⭐ The west-side duck at the ladder entry and the ports is exactly as accepted: `9_TH_EAVE_W` §13.4 “THE MEMBER OF RECORD” — single rough true 4×4, ruled clear span 26.25 in, utilization 0.270, governed by SHEAR (NDS 3.4.2), PASS, 61.9 % of capacity unused. Its run-cell note records that 1.2e is why the run re-based from 53.75, and that the two ports (~16.0 and ~11.5 in) are both shorter than 26.25, so neither governs.
⭐⭐ And that number cannot move with this proposal	The west eave check’s station-16 input (`eave__sta_deck_start`) is orphaned — after the R24/R25 re-base moved the run’s north end to the N double trimmer at 43.5, no formula on that tab consumes station 16 any more. So deleting members at station 16 cannot touch the 0.270. That is a verified fact about the live workbook, not an argument.
The EAST eave plate over the east porch segment	Also unchanged — and worth stating precisely, because the ring study’s finding ①②③ on this edge is easy to misread. The ring gave the east porch segment a posted NORTH end only (post 1, the NE corner, at station 16); its south end is still R21's unruled T-intersection. The proposal keeps post 1 — it is a corner post, not a plate — so that support story is identical before and after. ⚠ DERIVED HERE: the clear run is 25.5 in (post-1 south face at station 18 under R34 center registration → the door-wall line at 43.5), comparable to the west’s ruled 26.25 — but the T-intersection at its south end is unruled, so it is not a settled span.
R21's door wall at station 43.5	Untouched. R21 stands. Its single rough 4×4 plate, its two segments, its two J14 ends, the door and its rough opening are all outside this proposal, which reaches only to station 16. The R20-held door-height question is neither answered nor disturbed here.
The ridge, its four posts, and their reactions	The N-GABLE ridge post stays continuous floor → ridge soffit (8'-11⅝″) at 127.9 lb, as at all four stations. The ridge beam still crosses the porch and is what you stand under at the ridge line. ⚠ One thing about that post does change — see the second finding below.
The wind sail, and the dead the workbook charges	Both already model this face as if no wall were there. `2_LD_WIND` §14.4: the E–W line of action sits at station 77.0, the centroid of a face running “station 16 to 138 because the ROOF carries north over the covered porch even though no cabin wall does.” And `2_LD_DEAD` §7 charges the north wall zero (“STRUCTURALLY FREE… NO wall dead charged, never-invent”). The sail model stays conservative; the dead model is simply true.

⑥ The headroom won

The cathedral ceiling over the porch is the rafter soffit plane — the plane through the birdsmouth seats — except across the ridge beam’s own 4-in footprint, where the 10-in beam hangs below the rafters and is the ceiling. That is 2.12 §9.2’s own headroom rule: you measure to the beam soffit, because wood hangs below the beam top. ⛔ Z_APEX 120 is a seat-line construction point and geometry.py forbids printing it as a soffit apex — the trap the R20-held door memo fell into. The drawn soffit apex is 9'-10" at X 68.46, and a build guard refuses the page if it is not below Z_APEX.

where you are standing	clear height	vs the deleted plate	what the ceiling is there
At the west wall line (X 0–4)	5'-2"	—	the W eave plate’s underside — unchanged, this is the wall
1 ft in from the west wall (X 16)	6'-4 5/8"	+1'-2 5/8"	rafter soffit
X 32	7'-5 1/4"	+2'-3 1/4"	rafter soffit
X 48	8'-5 7/8"	+3'-3 7/8"	rafter soffit
⭐ At the ridge line (X 66.5–70.5)	8'-11 5/8"	+3'-9 5/8"	the ridge beam soffit — the beam hangs 10 in below the rafters and IS the ceiling here (2.12 §9.2)
X 88	7'-7 5/16"	+2'-5 5/16"	rafter soffit
At the east wall line (X 104–108)	5'-2"	—	the E eave plate’s underside — unchanged

The band that matters. Solving the soffit planes for a given clear height — DERIVED HERE, entered in no source:

6'-0″ clear from X 10.11 to X 102.11 — a band 7'-8" wide.
6'-8″ clear from X 20.26 to X 96.26 — 6'-4" wide.
Under a plate at Z 66 both bands are zero: 5'-2" everywhere on the face.

⭐ The east edge of the 6-ft band lands at X 102.11. The frozen north-knee-frame study derived the same X 102.11 independently as the east jamb of a 6'-0″ door head — two studies, two questions, one plane.

⑦ Enclosure — one line, and it is the frozen sibling's

Under the owner’s stated reading 1 (the envelope closes at station 43.5, so the covered porch bay is outside it — his working assumption, carried as a PE-confirm line into brief v3), the openness of the station-16 face appears in no term of the ASCE 7-16 §26.2 test — not A_o, not A_oi, not A_g, not A_gi. That is unchanged from the frozen triangle study, which worked the test on both live readings; nothing on this page re-opens or re-computes it, and no classification is claimed. Under reading 2 the aperture can only grow, so the proposal lands on the same side of that study’s condition-(1) straddle as its options (a) and (b) — which is exactly why the study handed the reading itself back as NEEDS_CONTEXT, PE/AHJ.

⑧ ⭐⭐ TWO CONSEQUENCES THE PROPOSAL DID NOT NAME

Finding 1 — the roof sheathing’s north edge loses the member it was going to be nailed to

2.6a §6.6 deliberately re-struck the rafter layout so that no rafter falls in a gable plane — the last rafter sits 5 in inboard of this line (4.625 in as tabulated on the drawn 121.25-in chord; R33-D flipped the basis of record to 122, which re-derives 5 — DERIVED HERE) so that both gable walls die into the sloped ceiling. The proposals memo is explicit about the consequence: the triangle’s top boundary “is therefore the roof sheathing plane itself, not an end-rafter soffit,” and the member that supports it is the raked cap on the gable studs — which is also, in P10, the roof diaphragm’s boundary member on the gable line.

⛔ This proposal deletes the studs, so it deletes the raked cap across the open middle of the face. What is left at that edge is the last rafter 5 in inboard, the 6" NOMINAL furring rake cantilevering past the line, and whatever barge or raked-cap treatment is chosen — and that treatment therefore becomes the entire diaphragm boundary story at this edge, where under P10 it was one member among several. P10 asks the PE for 8d at 6 in on boundaries plus 4 in o.c. within 48 in of roof edges and at gable-end framing, on this workbook’s own adopted Exposure C. That nailing needs something to go into.

⛔ NEEDS_CONTEXT — PE. No source on file specifies the replacement, and none is invented here: not the member, not its span (the same 5'-6 1/2" / 3'-1 1/2" the deleted plate had, now sloped and with only the corner strips under it), not its bearing, not its nailing. Figure 2 says so on its own face. The related standing gaps are already the workbook’s: §12h item (4) leaves the chords “by default”, §14.7 records the porch-region edges as “UNCHECKED and UNNAMED”, and §14.8 records that plate-to-stud uplift ties have no device, fastener or schedule specified.

Finding 2 — the N-GABLE ridge post loses the only member that ever framed into it at mid-height

9_TH_POST_RIDGE takes each ridge post’s length as “the LOCKED plate height 5'-6″” — an admitted proxy, recorded as such in geometry.py’s datum v6 rule D. The frozen post-bases study quotes the consequence: l_e/d 16.5, axial 0.094, bearing governs, not buckling.

The proxy predates the ring, so the ring did not create it — but the ring did make it true, at both gable lines: J14 puts a plate end into the post’s side at exactly Z 66. ⭐ Deleting the north segments withdraws that member, and station 16 becomes the one gable post whose modeled length has nothing behind it. Geometrically the unbraced height in the E–W direction goes from the modeled 66 in to the post’s full 107.635 in — l_e/d 16.5 → 26.91, DERIVED HERE and nothing more than arithmetic on RIDGE_POST_FLOOR_TO_SOFFIT and POST_W.

⛔ No capacity is computed on this page and none may be read into it. Two things are worth saying alongside it, both of which point the same way: this is the lightest of the four ridge stations (127.9 lb against 959.9 at the south mid), and the corner strips — whatever their ruled width — do not reach the ridge post, which stands in the open middle. So this reads as a margin question, not an alarm. It is a Domain B row to re-run on the honest length and a PE item, and it belongs on the v11 list.

⑨ If adopted — the v11 delta list

⛔ Nothing below is done, and this page does not do it. It is what a Fable datum pass would have to write, and it is listed so the owner can see the size of the ruling before making it. The datum is v10 and stays v10 until a Fable pass writes v11.

what moves	the delta
Rule A — “four plates close a ring”	The ring becomes a U: two eave plates + the south gable line’s two segments. `GABLE_PLATE_SEG_W/_E` survive on the south line only. The rule’s own closure assertion still holds (it is about the split, not about how many lines carry it), but the word ring stops being true.
Rule A — the six R30/R31 posts	Still six. The NW and NE posts change job: they carry an eave plate END rather than a plate corner. ⛔ Their top condition is OPEN — see the open items.
Rule A — J13	Four gable-line corners → two (south). The derived lap handedness and the HTP37Z alternate are unchanged for those two.
Rule A / rule D — J14	Six joints → four. Same single application, two fewer instances, same `PE TO SPECIFY`.
A NEW joint, currently scheduled nowhere	Eave-plate END → corner-post TOP, two instances, at the NW and NE corners. Under the ring this joint was J13’s half-lap; without a perpendicular plate it is a different joint with different hardware. It needs its own J-row.
Rule B — R34 registration	Unchanged for the posts (they still center on the as-built north nailer). The half of rule B that registers a gable plate to that nailer has nothing left to register at station 16.
Rule C — the 6-in rake	The rake itself survives (it is a roof-edge member, framed by extending the furring). ⛔ What does not survive is the member the roof sheathing’s north edge was going to be nailed to. See finding 1.
Rule G — the open/queued list	The line “the N plate’s beam duty over the open porch face (clear spans 62.5 / 33.5) — Domain B calc row” is withdrawn. The line “gable framing ABOVE the ring plates (both lines)” is answered on the north line — no wall — and unchanged on the south.
Stock & the BOM	One 12-ft rough 4×4 stick is released: R33-A’s four-stick ring becomes three. Domain B owns row 129 and the cut list; no figure is priced here.
The construction set	Any sheet that draws the north gable line at Z 66 would be re-rendered by the same Fable pass. Nothing in the set is touched by this study.

⑩ Open, unsourced or handed back — raised, not fixed

⛔ Nothing was corrected in any source from here. The never-invent rule governs every item: where the ladder of sources ran out, the item is named and left open rather than drawn.

⭐⭐ The CORNER-STRIP EXTENT — the widest open item on the page — The owner’s words are “associated supporting NW/NE 16-oc framing up to the rafters.” No source states how wide that strip is — how many studs, to what station inboard, or what it is holding up. Figure 1 draws the corner post plus one stud SOLID and a second DASHED purely so the reader can see what a strip looks like; the drawn width is illustrative and carries no authority. OWNER TO RULE, and the answer changes what the middle of the face means.
⭐ The NW / NE CORNER POSTS’ TOP CONDITION — The proposal says the corner posts run “full height to the eave plate ends / roof edge, not stopping at 62.” That admits two readings and no source settles either: (i) the post tops at the plate underside, Z 62, with the eave plate running continuous over it — the established R28/R30 condition minus the perpendicular member, which is what Figures 1 and 2 draw; or (ii) the post continues past the plate to the roof plane, which needs the plate notched, let in, or side-framed. OWNER TO RULE — and (ii) would change the eave plate’s own bearing, which is R5/R27 territory.
⭐ The PLATE-END-to-CORNER-POST HARDWARE — Named by no source, and not inferable from J13 (a half-lap is not a post cap) or from J14 (that is a plate end into a post side, not a post top). ⛔ The owner has separately proposed Simpson ROUGH-CUT post caps — the AC4RZ / BC4R family — which would be exactly this joint’s hardware. HARDWARE UNDER VERIFICATION — companion research pass. Nothing is claimed for it here: no literature for that family is on file in docs/product_specs/, and it has no row in docs/components/. Per the registry’s own rule, absence means “not yet seeded”, never “no ruling exists” — and it is never a reason to draw a part.
⭐⭐ NEEDS_CONTEXT — the roof-sheathing boundary at this gable line (finding 1) — See finding 1 above. PE, and it is the sharpest thing on this page.
⭐ The N-GABLE ridge post’s unbraced length (finding 2) — See finding 2 above. Domain B row + PE.
The east eave plate has never been checked, at any span — Raised while sourcing, and it long predates this proposal: there is no 9_TH_EAVE_E tab. 9_TH_SCHED’s 40 instance rows carry exactly one plate member, the west one. The east plate’s 25.5-in porch run is almost certainly the least of its worries, but nothing anywhere says so. Domain B.
The porch west wall is still not counted, and nobody has ruled its sheathing — 9_TH_EAVE_W carries the sentence three times: “THE PORCH WALL IS NOT COUNTED IN Sigma-b — R8 adds candidate length north of the trimmer but does NOT rule sheathing for it, and that is a separate question nobody has answered.” Unchanged by this proposal, and named so a reader does not assume the open porch removed it.
The rake is NOMINAL — 6 in at both gables, v10 rule C, with [PE/MFR] max panel rake overhang + rake trim detail FIELD VERIFY AT ORDER still open. This proposal makes the north rake’s edge detail matter more, not less.
R34 registration is FIELD-PENDING — The posts center on the as-built nailers; the north nailer strays ~1.22 in north at the west end (NAILER_N_STRAY_W_DERIVED). FIELD VERIFY — tape the nailer faces. Nothing here is blocked on it.

⚠ ERRATUM FOUND WHILE SOURCING — RAISED, NOT FIXED (discrepancy protocol). In 2_LD_WIND, the Σb notes disagree with the live cells they sit beside. The notes at the per-direction and per-line Σb rows state 13.5417 ft (E–W) and 11.0625 ft (N–S); the live cells recalculate to 11.2917 ft E–W (81 + 54.5 in, after R18 netted the SE-3 window) and 10.2292 ft N–S (54.5 + 68.25 in, after R19 set the east window width to 40 in). The N–S torsion note likewise reads 346 lb-ft against a live 855.6 lb-ft. Quote the cells, not the notes. Dated correction owed to Domain B; none of it changes anything on this page, which quotes cells throughout. ⛔ Nothing was changed here.

Deliberately NOT drawn

A recommendation · a ruling · any selection between the proposal and the v10 ring · the corner-strip width · the corner posts’ top condition · the roof-edge boundary member, its span, its bearing or its nailing · a connector, part number, size, count or capacity for the plate-end-to-post joint or for anything else (the AC4RZ / BC4R family is named as under verification and nothing more) · a utilization, a reaction or an allowable for any member · an enclosure classification · a price · a J-row · a component-registry ruling · a datum edit of any kind · any correction to calcs.xlsx, geometry.py, the proposals memo or any register · any modification of the three frozen sibling studies, which are quoted only.

Guards

build_study.py refuses to write the page if any of these fail, and names the offender:

⭐ the datum gate — geometry.py’s banner must still read DATUM v10, read off the banner itself
⭐ the premise gate — PLATE_W = PLATE_DEPTH = POST_W = 4.0, PLATE_STEP = 0
⭐ the thing being deleted must still exist — the gable line must close, 66.5 + 4 + 37.5 = 108 exactly, and the two clear spans must still be 62.5 and 33.5, or this page is about nothing
the members that remain must remain: the ridge post continuous past the plate at all four stations, the eave plate length model untouched, and the west ladder entry still at 26.25 in ≥ the 24-in allowance
⛔ the drawn soffit apex must sit below Z_APEX, which geometry.py forbids drawing as a soffit apex
⭐⭐ the ceiling model — the rafter soffit must be above the ridge-beam soffit at both beam faces (or “the beam is what you stand under at the ridge” is false), the ceiling must rise monotonically from each eave toward the beam, and the printed gain must equal beam soffit less plate underside exactly
⭐ the headline must still be a headline — the plate underside must still sit below 6 ft (so “nowhere on the face clears six feet today” is true), the 6-ft band must land inside the wall span, and a higher clearance must buy a narrower band. If any of that stops holding the page is rewritten, not re-run
⭐ the roof edge — no rafter may fall in the gable plane (2.6a §6.6's re-strike is what makes finding 1 real), the inset must derive 4.625 on the drawn chord and 5 on the ruled basis, and the rake must reach past it
⭐ the slenderness fact — the full post must be more slender than the modeled proxy, and the proxy must still be l_e/d 16.5; the guard is on the geometry only, because no capacity is computed
⭐ handedness — all three figures put EAST on the right, deliberately, so they read together: Figure 1 looks NORTH from inside the porch (an exterior north elevation would be its mirror — the A-201 view-direction ambiguity, stated rather than assumed), Figure 2 looks NORTH, Figure 3 is a plan from above with north up; and up is up in both elevations
every SVG label measured against its own viewBox, horizontally and vertically, on a deliberately generous character-width table; every wrapped caption measured to a line budget so nothing breaks mid-word
a forbidden-claim scan (no adoption, no selection, no capacity, no classification, no connector named as chosen, and v11 may never be asserted as a fact) and a required-phrase scan (PROPOSED · FIELD VERIFY · PLANNING — NOT FOR CONSTRUCTION · the datum · the freeze rule · DERIVED HERE · never-invent · NEEDS_CONTEXT · Domain B · R21 · R28 · R30 · R32 · R34 · P1 · P2 · P10 · J13 · J14 · 2.6a · 1.2e · 9_TH_EAVE_W · owner rules)

STUDY — illustrative, frozen at DATUM v10 (2026-08-14), generated 2026-08-15 from geometry.py and a READ-ONLY reading of calcs.xlsx. PLANNING — NOT FOR CONSTRUCTION. The proposal is PROPOSED; the owner rules. ⛔ Do not re-run this generator on a later datum — write a new study.

PLANNING — NOT FOR CONSTRUCTION — PE REVIEW PENDING
Generated from geometry.py — datum v11 · 2026-08-15 · set dated 2026-08-17 · 13 sheets. Every sheet is regenerated whenever the model changes; unlike the studies, a stale sheet here is a defect.
Basis: Pg 40 · Ce 1.2 · Ct 1.2 · Cs 0.871 · E. Hem-Tam No.2 (culled) · wet-service

A-901 · ILLUSTRATIVE PERSPECTIVE

S-001 · GENERAL NOTES & SCHEDULES

S-101 · ROOF FRAMING PLAN

S-102 · FOUNDATION & PILE PLAN

S-103 · DECK FRAMING PLAN

S-104 · WEST-MID SUPPORT — OPTIONS

S-201 · S GABLE — FRAMING ELEVATION — LOOKING NORTH

S-202 · PLATFORM & SUPPORT ELEVATIONS

S-203 · RIDGE SUPPORT DIAGRAM

S-204 · RIDGE SUPPORT PLAN

S-301 · SECTIONS & DETAILS

S-302 · CHASE COLUMN & BEARING DETAIL

A-201 · S GABLE — FINISHED ENVELOPE ELEVATION — LOOKING NORTH

Renderings answer “what will it be like?” — the third drawing class (adopted 2026-08-16). Like the set they are regenerated with the datum — a stale rendering here is a defect, not a snapshot — and like A-901 they are illustrative and must never be measured from: no dimension string exists anywhere in them, by build-time guard. Generated on datum v11 · 2026-08-15. PLANNING — NOT FOR CONSTRUCTION — PE REVIEW PENDING.
Where structure reads as a band on the outside — the plate line on the south and east faces, the ridge-post board on the south gable — that is applied, flashed trim over the envelope (R46), not exposed structure: the plate and the ridge post stay behind continuous sheathing.

⭐ The Experience v1 — the whole structure, real-time. Drag to orbit, scroll to zoom, shift-drag to pan; the nine buttons are the preset cameras. WALK MODE switches to first person — W A S D to move, drag to look, Q/E for height, hold shift to move faster.

The flat set · eight views · 01→08

01 · SOUTH — the gable the windows are ruled into

02 · EAST — the one window, and the envelope it is ruled into

03 · NORTH — the cathedral face

04 · WEST — the arrival side, and the tree coming through

05 · THE PORCH — the cathedral volume, looking in

06 · THE CABIN — limewash, exposed frame, and the timber band

07 · THE LOFT — at eye level, on the net

08 · THE CHASE WALL — the plexi, and the oak behind it