Build Addendum: Dimensional Lumber Adjustments & Foundation Decision

10x12 Gable Shed — Gibson St, Atlanta GA

Prepared July 2026, verified line-by-line against PLANS.GBL.10X12.SHELL.pdf. Updated Oct 1, 2026 (roof: SYP rafters and ridge, connectors, asphalt roof; west window centers).

---

PART 1 — Building the Plans with Store-Bought (Dimensional) Lumber

The core fact

The plans are dimensioned for rough-sawn full-dimension lumber (a "2x6" that is actually 2" × 6"). This is stated three times:

• Page 1: "We use rough-cut full dimension hemlock for our framing... A ¼″ difference is not uncommon."
• Page 3: "The plans are set for full dimensional lumber."
• Page 40 (Appendix A): "Rafter lengths supplied are set up for full dimensional lumber" and "Rafter templates have been set up for a one inch ridge."

Store lumber is ½" smaller in each cross-section direction:

Note: the plans' own cut list assumes rims ~1¾" thick (floor joists 116½" + two rims = 120" only if rims are 1¾"), so even the JCS kit varies. The design is rough carpentry meant to be fitted in place, and it absorbs dimensional lumber fine. The danger is only in pre-cutting from the printed cut list.

The five golden rules

1. Never batch-cut the whole cut list up front. Cut primary members to the printed lengths; cut every infill piece to measured fit.
2. Absolute lengths stand as printed — they set the 120" × 144" footprint and are independent of lumber thickness: skids 144" (±¼" — nothing mates to a skid end), rims 144", top plate beams 144"/120", posts 74" (4x4), door header 62½" (1x4), ridge 1x8 × 144" (a full 12' board, uncut) — Superseded Oct 1, 2026: ridge is one 2×8×12 #2 KD SYP cut to 143-13/16″, angle braces 4x4 × 30" long-point-to-long-point.
3. Anything that fills between two members is wrong by ¾"–1" (see table below).
4. Hold the rough openings, not the printed layout stations. Door RO 62½" × 73", window RO 24½" × 24½". Set the inner faces of door posts and window studs to preserve the RO; layout marks in the plans assume 4"-wide posts. (Page 31: RO may vary ±¼".) This applies vertically too: the flat-laid window sill nailer and header are 1½" thick, not the assumed 2" — register the RO-forming edges (nailer TOP at 36", header BOTTOM at 60½") and let the far edges land where the thinner stock puts them. The finished door opening moves too: the plans show 60½" × 72" after the 1x5 jamb trim, but that is computed on JCS's full 1"-thick 1x stock. Our ¾" jamb gives ~61" wide × ~72½" high. Build the doors to the opening you actually measure once the jamb is in — never to the printed number — and allow the plans' 5/8" gap (p.31).
5. The deck is the reference. Keep it exactly 120" × 144", diagonals 187½". Everything above is cut to the actual building.

Cut-length conversion table

Window coordinates superseded Oct 1, 2026. The Aug 25 stud coordinates and centers in this section are superseded by the Sep 2 field-confirmed west window stud centers: 25-3/4″, 61-1/4″, 82-5/8″, 118-1/8″ from the south outside face (clear RO width 34″). Use Wall_Framing.html (and Window_Install_and_Drainage.html). The brace history below still stands. W2/W4 openings are still solid sheathing (not cut out yet).

As-built, 25–26 Aug 2026 — two west-wall angle braces, and the west windows moved 3". An extra 4x4 brace was installed by mistake at the NW corner of the WEST wall, a wall with windows, which the plans exclude. It was left in place deliberately — record this as an as-built deviation, not an error to fix. Removing it would mean prying six 16d toenails at a plumbed corner, which risks racking the frame — the same mechanism that put 3/8" of rack into the plate box. It is also not harmful: the plans exclude braces from opening walls because in the stock design the siding provides those walls' shear and a corner diagonal would cross their 66" window headers. Our plywood sheathing shears every wall regardless, and our headers are only 34" and sit higher. The brace is redundancy, not a problem. Consequence: eight braces installed, six per the plans — six pieces were cut from two 4x4x8' PT boards, so the west-wall brace spends a third (+1 board, ~$16). The 72" girt row at that NW corner now runs as a ~12½" stub, the same as the braced corners on east and north.

The windows moved with it. Both west-wall windows shifted 3" toward the center of the wall so the rough openings clear the brace — as originally laid out the clearance was only 0.34", inside both the cutting tolerance and the plans' ±¼" RO forgiveness (p.31). As-built ROs 22.84"–56.84" (window 1, south) and 87.03"–121.03" (window 2, north) (superseded, see above); studs at 21.34–22.84 / 56.84–58.34 / 85.53–87.03 / 121.03–122.53, centers 22.09 / 57.59 / 86.28 / 121.78 (superseded: Sep 2 centers 25-3/4″ / 61-1/4″ / 82-5/8″ / 118-1/8″); both still symmetric about the wall center at 71 15/16". This was a horizontal move only — sill height unchanged at 48", head unchanged at 67⅝". Cut-list detail: the north window stud stops at 71", not the full 74", dying into the brace; the brace backs the sheathing seam above it, so nothing is lost. (Superseded Sep 2: all four window studs run full height.) The change pushed exterior sheathing from 11 sheets to 12. The trade-off: from inside, the original layout had four equal 14.8" gaps between verticals and the new one reads 17.8 / 11.8 / 11.8 / 17.8; from outside, where the center post is invisible, the three visible spans become more even — 19.8 / 36.2 / 19.8 becomes 22.8 / 30.2 / 22.8. The west wall faces the yard, so the exterior read was chosen. Full stud-by-stud layout in Wall_Framing.md (current) / Wall_Framing.html.

Knock-on dimensions

• Wall height: 77½", not 78" (74" post + 3½" top beam). All heights above the plate drop ½". Overall building height ends ~1½" shorter than the listed 10'6" (thinner floor stack + thinner beam). Nothing depends on the printed heights — measure the building.
• Door RO height becomes ~73¼" (74" post − ¾" header instead of 1"). Harmless; doors are built to fit after framing.
• Floor sits ~1" lower (3½" skid + 5½" joist vs 4" + 6"). Ramp angle marginally shallower. Fine.

Rafters — the one careful spot

Species: 2×6×12 #2 Prime KD Southern Yellow Pine (supersedes the Aug 11, 2026 spruce decision). Not PT, not cedar (cedar 2x6x12 is stocked only GREEN and checks and twists), and spruce is out. Connectors are decided and listed in the roof-seat audit follow-up below (H2.5AZ at the interior seats, A35Z at the end seats and ridge ends, RR-R at the ridge; all with SD9112 screws).

Do not trace the paper templates. They are drawn for a 6"-deep rafter and a 1" ridge (page 40 note). Instead:

1. Lay out with a speed square: 6/12 pitch (26.5° plumb cut).
2. Top-edge length: 66-1/4″, plumb cut to plumb cut (about 68-1/4″ overall, tip to tip), determined from the test pair. (66-5/8″ was the JCS original — a different number; it was a reference, not the cut length.) A 72″ half-board is enough for one rafter.
3. Lay out the JCS end-seat cut: level bearing at the rafter end, plumb outer end against raised sheathing, no projecting tail. Measured (Oct 1, 2026): plumb end against the wall sheathing plus a level seat, heel height 4-3/8″, seat ~3-1/2″, no tail; custom seats or shims per plate height. The ~3-1/2″ seat happens to equal the old withdrawn rule's number, but it is now field-derived from the test assembly, not the old unverified rule.
4. Include actual ridge thickness in the test-pair layout; do not dismiss the change from the original 1″ ridge.
5. Cut ONE TEST PAIR first. Stand them on the actual building with a ridge offcut sandwiched at the peak. Verify peak closes and seats sit flush, adjust, then use the proven pair as the pattern for the remaining twelve. (The plans themselves prescribe cut-one-test-one.) Status Oct 1, 2026: done — the test pair was fitted and all 14 rafters are cut (one miscut; a replacement 2×6×12 is being bought Oct 2). Raising is planned for Oct 2.

Ridge check: dimensional 1x8 (7¼") still fully backs the rafter plumb-cut face (~6⅛" on a 5½"-deep rafter). ✔ Superseded Oct 1, 2026: the ridge is a 2×8×12 #2 KD SYP cut to 143-13/16″ (top 34″ above the plate top, bottom 26-3/4″), 1½″ thick, with RR-R brackets attached.

Roof-seat audit and connector fit — 19 September 2026

Status Oct 1, 2026: resolved. The roof seat and connectors are decided (see below); the audit is kept as history.

Decision: preserve the JCS rafter end-seat shape. A level bearing cut at the rafter end rests on the plate; the plumb outer end abuts the raised bearing-wall sheathing. There is no projecting rafter tail and no new internal tail notch.

Provenance: JCS Appendix A, PDF p.40, uses generic synonymous terminology for its seat cut. PDF p.43 supplies the actual end-seat template, and p.28 shows the installed no-tail geometry. The archived addendum and build guide already repeat the generic terminology and introduce the fixed seat-length/remaining-depth instruction; the Aug.11 species revision repeats it. No explicit decision to change the rafter profile was found in the project Markdown, active HTML, archived HTML or planning summary. No version-control history or complete prior LLM transcript was available, so the first author/date and any undocumented conversation cannot be established. The original purchased PDF and embedded source-page images are preserved unchanged. Derived prose, including archived copies, has been corrected.

Cutting correction: the previous 3½″ seat / at-least-4″ remaining-depth rule is withdrawn as an unverified translation, not retained under a new name. (Oct 1, 2026: the as-cut length is 66-1/4″ top edge, plumb cut to plumb cut, about 68-1/4″ overall; 66-5/8″ was the JCS original.) The original 66⅝″ length was not final modified-roof cut authority. Derive the test pair using actual plate spacing, rafter depth, ridge thickness and end-seat geometry. Do not change the roof profile just to fit hardware.

H2.5AZ fit assessment — not a blanket approval of fourteen identical installations:

• Ten intermediate seats: the JCS end-seat shape does not inherently prevent an interior-mounted H2.5A-type connector. Its upper flange fastens to a broad rafter face, and its lower flange to the exposed inside face of the bearing plate, with the stamped plate line aligned as instructed. Orient it so the rafter flange lies toward the ridge, not toward the absent tail. The 3½″ plate is deeper than the connector's approximately 2 3/16″ lower portion. The approximately 3 13/16″ upper portion can fit beside a nominal 2×6 at the inner bearing edge. This is a geometric screening result, not measured verification of every hole in an as-yet-uncut rafter. Check the actual connector against the test assembly before fastening. Decided Oct 2026: H2.5AZ ×10 at the 10 interior seats (installed as each rafter is set — raising Oct 2); RR-R ridge connectors on the 10 interior rafters at the ridge (RR-R ×10 used; 4 extra bought are to be returned).
• Four end seats: the end rafters are flush with the building ends, while the gable plates extend across the building and half-lap the bearing plates. The gable plate occupies the interior corner space where the ordinary interior tie's lower flange would sit. The uninterrupted-face assumption used for the ten intermediate seats does not hold here. Do not bury a flange in the lap, attach through an air gap, improvise bends, omit holes or notch the structural plate to make it fit. A different approved placement/connector detail must be resolved for these four seats. Resolved Oct 1, 2026: A35Z ×4 at the end seats — installed; A35Z ×4 at the ridge ends (the RR-R does not fit at the ends) — to be bought Oct 2. The cautions above stand (do not bury a flange in the lap, no improvised bends). The blanket fourteen-identical-ties purchase/install instruction remains withdrawn.
• Fasteners: SD9112 #9 × 1½″ for all connectors (no hand-nailed connectors). Per Simpson's table for the H2.5A installation: five Simpson SD9112 #9 × 1½″ connector screws into the rafter and five into the plate, per Simpson's U.S. connector table. These are not SDWS framing screws. Ten intermediate ties use 100 screws; total need ≈ 276 (RR-R 10×8 = 80, A35Z 8×12 = 96, H2.5AZ 10×10 = 100); 3 boxes on hand + 1 box Oct 2. Use the appropriate corrosion protection for PT contact.
• Connector fit does not establish uplift demand, plate-to-post anchorage or the rest of the load path. No whole-building uplift approval is implied. Temporary bracing remains until the original JCS removal checkpoint.

Sources: original JCS PDF pp.28, 40, 43; Simpson H2.5A installation guide; Simpson approved SD connector screws; Simpson H2.5A CAD drawing (geometry only; use U.S. fastener/load specifications, not UK ratings).

Structural impact of nominal lumber (checked)

Nominal lumber has ~63% of the bending strength and ~58% of the stiffness of true full-dimension stock of the same species. Offsets: PT southern yellow pine is a stronger species than eastern hemlock, and every span in this building is short.

• Floor: joists clear-span only ~7'5" between skids (skids inset 10" per plans). Nominal PT 2x6 @ 24" o.c. is adequate at residential floor loading, slightly bouncier than the hemlock kit. Recommended upgrade: 16" o.c. (+4 joists, ~$50) if a riding mower or other heavy point loads will live inside. Cheapest insurance in the project.
• Roof: rafters span ~5' horizontal per side; Atlanta snow load is negligible. No change needed — but the roof is now asphalt, about 1,160 lb installed; see Structural_Load_and_Connection_Assessment.md §11.
• Walls/posts/beams: loafing. No change needed.
• Doors (Appendix B): built from two layers of boards; ¾" stock yields a 1½"-thick door vs JCS's 2". Functionally fine. Use 5/4 boards for the face layer if you want the heft.
• Fasteners: unchanged (16d framing, 16d galv exposed 2x, 6d galv for 1x — page 9 guide still applies).

Integration fix for the cedar-shingle modification (our change, not the plans')

The plywood sheathing added in the Feb 2026 cedar estimate does not line up with this frame: posts are 72" o.c., the one horizontal nailer is at 34–36", so 4x8 sheet edges land on nothing in either orientation.

• Add flat 2x4 blocking behind all unbacked plywood edges — roughly six 2x4x8 PT (~$45). (Historical — walls sheathed Oct 1; not tracked.) Not in the Feb estimate (its "no furring strips needed" note covered shingle nailing, not seam backing).
• Run sheathing up to the underside of the roof strapping; fascia/shadow board close the top as in the plans. Superseded Oct 1, 2026: there is no strapping. Sheathing was finished Oct 1: north and south sheathing was cut back to mid-plate (the gable plywood strips nail to the plate's upper half); east/west sheathing projects ~1-3/4″ above the plate. The shadow board sits at the TOP of the fascia — a deliberate departure from JCS p.30, which centers it. The wall plane sits 15/32″ (ply) vs 1" (pine boards) from the frame — rafter ends and trim reveals absorb this. The wall weather barrier is #30 felt.
• Bonus: plywood adds racking strength the pine siding was providing; keep the 4x4 angle braces anyway (cheap, and they're part of the interior look). This is also why the two as-built west-wall braces are harmless — the sheathing, not the brace pattern, is what shears each wall.

Shopping list tweaks vs Feb 2026 CEDAR estimate

---

PART 2 — Foundation: Gravel + Solid Blocks vs Poured Piers

Decision: gravel bed + six solid concrete blocks (as planned)

This is JCS's primary recommendation (pages 16, 46: blocks are how "the majority of our buildings are installed") and it is the right choice for this building in Atlanta.

Why blocks beat shallow piers here

1. "Self-leveling concrete" is not a thing for footings — it's an interior floor-topping compound. What a pier approach actually requires is setting six tube forms to a laser line and screeding six separate pier tops into one flat plane across an 8'×11' rectangle. Wet concrete levels only within its own form, never across forms. That is harder than leveling blocks, not easier.
2. Asymmetric failure modes. Miss with blocks → add a shim in 30 seconds. Miss with concrete → grind cured concrete, or shim on top of the piers anyway. Blocks are adjustable forever; concrete is adjustable never.
3. Atlanta clay moves. Piedmont red clay swells/shrinks seasonally; the moisture-active zone runs several feet deep, so shallow piers don't escape it — they just remove adjustability. (The plans' "dig 48 inches" is Vermont frost-line advice; Atlanta frost depth is <12".) Expect some small differential movement over the years regardless of support type; the skid-on-block design exists precisely so the building can be re-leveled with a jack and shims in an afternoon.
4. Loads are trivial. ~4,000 lb building + contents over six 8"×16" bearing faces ≈ ~1,000 psf on compacted gravel — far below clay bearing capacity.

When piers would be right: site slope > ~6–8", engineered anchorage requirements, or a permit reviewer demanding footings. None apply to a flat backyard install.

Pad recipe (this determines whether it stays level — spend the effort here)

1. Strip sod/organics from a 12' × 14' footprint (1' beyond building all around).
2. Landscape fabric over the clay so stone doesn't migrate into it.
3. 4" crusher run (GAB), compacted in two lifts — plans specifically say compacted, not loose. ≈3 tons, ~$150–300 delivered bulk. Plate compactor rents ~$90/day.
4. Laser/transit-level the pad before blocks. Also level the ground ~5' in front of the door (plans requirement — the ramp must sit on ground level with the building).
5. Six solid 4"×8"×16" cap blocks (never hollow-core) in the rough 8'×11' layout, page 16 / pier diagram page 19 (ends 11"–13" in, ~6' spacing along skids). Bring tops to plane with the laser; final tune with shims under skids. If the grade needs more height on one side, stack solid blocks (max ~2 high dry).
6. Level the deck frame before the plywood goes down (plans, page 25 step 1: the whole building's squareness depends on it).

Atlanta-specific add-ons (undersold by the Vermont-oriented plans)

• Termite shield (Appendix C detail): aluminum flashing between blocks and skids, extending 2" out and bent down 45°. A throwaway line for Vermont; genuinely worth it in Georgia. A few dollars.
• Wind anchors: 2–4 earth-auger anchors with straps over the skids (~$20 ea). The shed's weight is nominally adequate per JCS, but cheap insurance against thunderstorm gusts under the asphalt roof.
• Drainage: the plans' #1 durability warning is water under the building. Crown the pad slightly; if the yard slopes toward the site, add a shallow swale or French drain uphill. Keep mulch/clutter away from the skirt for airflow (and termite visibility).
• Window rough-opening flashing: a sill pan is a shallow waterproof tray under the window that catches the water that gets past the sash and sends it back outside. It does not stop the leak; it makes the leak harmless. Water always gets past a sash eventually — wind-driven rain, cracked glazing putty, winter condensation running down the inside of the glass. With no pan under it, that water soaks into bare framing and rot starts in the two bottom corners, hidden behind trim. Three features are required: a back dam, a lip turned up at the inside edge so water cannot run into the room or the wall cavity; end dams, the pan running 4–6" up each jamb so water cannot run sideways into the corners — miss these and you have built a bathtub that drains into the wall; and open and sloped at the front, so water has exactly one way out, over the face of the sheathing. Cost is ~$25 for a roll of 4–6" self-adhered flashing tape, plus ~$18 for housewrap seam tape — and the seam tape is genuinely optional, since cap nails hold the wrap on but do not seal the laps, and plenty of outbuildings get 6" laps and nothing else. This matters more on this build than on a normal one: the two salvage barn sashes going into these openings have alligatored paint, failed glazing putty, an open corner mortise-and-tenon and one punky bottom rail. They will pass water this winter, before restoration. Full write-up, including the lapping order, in Wall_Framing.html under "Windows and weather — the rough openings".

Cost

Poured piers would run similar-or-more (tubes, ~7 bags concrete, digging 6 holes in clay) for a less serviceable result.

Site-specific drainage plan (added July 2026, after pad siting)

SUPERSEDED (late July 2026): the design below predates the site-orientation and oak-root corrections. The current as-built drainage design (collector on the east/south toes south of the W→E root, SW daylight exit, geotextile floor separation) lives in Site_and_Foundation_As_Built.html, and in Shed_Build_Guide.html Phase 1. This section is kept for the decision history only.

As-built site: pad benched into the high corner of the lot (cut up to a few feet deep at the fence side) with ~1 ft between fence and the 12x14 pad edge (+4x6 timber edging). Own yard drains away; only inflow risk is runoff from the slightly-higher neighbor to the south, plus direct rain and roof water. The recess is effectively a clay bathtub — the fix is an outlet, not uphill interception.

Priority order:

1. Drain the tub to daylight (essential). The excavation floor is level — that's fine; water seeks its own level and only needs an outlet below floor level (bathtub principle). Dig a narrow collector trench along the toe of the cut face and/or downhill edge, 3–4" deeper than the pad subgrade, trench bottom falling ~⅛–¼"/ft toward the exit corner. 4" perforated pipe in fabric-wrapped gravel (or continuous gravel trench), then out through the downhill lip until grade falls below the pipe invert (~10–20 ft near the hilltop). Water table in the pad then can't rise into the gravel at all. Retrofittable after gravel placement — perimeter dig only; max lateral travel through gravel is ~7 ft.
2. Toe-of-cut drain (cheap while trenching). Extend the same pipe along the inside base of the cut face (fence side) before it turns downhill; catches anything coming over/through the face.
3. Gutter the fence-side eave if a 12' eave wall faces the fence (half the roof ≈ 100 gal per 2" storm otherwise lands in the recess). Discharge at the downhill corner past the pad.
4. Storm test the fence line. If water visibly sheets under the fence in a hard rain, add a narrow gravel channel along the 12" fence strip, graded to carry flow laterally around the recess. If merely damp, skip — the toe drain covers it.
5. Watch the cut face. Clay will slough over wet/dry cycles — toward the pad and potentially undermining neighbor fence posts. A low stacked-block/timber facing against the cut stabilizes it.
6. Keep the shed-to-fence slot dry and clear. Stone the strip, keep leaves out, rely on the termite shield — dead-air damp slot is the prime rot/termite zone and it's near-invisible once built.

Anchors note: the recessed hilltop-corner siting is more wind-sheltered than an open yard; anchors downgrade to optional (2 on the exposed side, ~$40, still cheap).

Permit note

The shed is exactly 120 sq ft — the common IRC accessory-structure permit-exemption threshold ("not exceeding 120 sq ft"), which is presumably why JCS sized it that way. Verify City of Atlanta's current exemption and, regardless, zoning setbacks for accessory structures before siting the pad.

---

Standing rules — the ones that prevent ruined lumber

Migrated Aug 14, 2026 from the retired Order_Plan_and_Build_Sequence, which cited them as coming from this addendum. They live here now.

1. Never batch-cut infill from the printed cut list. Absolute lengths stand (144 / 120 / 74 / 62½ / 30); everything between members is +¾"–1" or cut in place.
2. Hold the ROs by their forming faces and edges, not printed stations. RO ±¼" is fine.
3. The deck is the reference: 120" × 144", diagonals 187½" (true 187.44 — chase equal). Tape beats print.
4. Rafters: test pair before batch. Templates assume full-dimension lumber and a 1" ridge — don't trace them. (Test pair done; all 14 cut as of Oct 1, 2026.)
5. Fasteners: HDG in PT, stainless on cedar, flush-not-sunk on shingles. Exception: the BHTOP 3/4″ electro-galvanized roof coil nails touch only the plywood deck and asphalt, never cedar or PT. Full rules in Nail_Guns_and_Fasteners.html.
6. Wall height as-built is 77½", not 78". Measure the building, not the drawings.
7. Angle braces: six per the plans, eight as-built. The extra one at the west wall's NW corner stays — deliberate deviation, documented above; don't "correct" it in the field.