Base & Deck Construction

Why our numbers differ from the plans

The single fact that drives everything on this page

JCS dimensions this building for full-dimension rough-sawn hemlock. We are building it from nominal pressure-treated stock. A 4x6 skid is the place that difference bites hardest, because a beam's bending capacity depends on the square of its depth:

We lost 30% of the skid's bending capacity to a lumber substitution that looks like it changes nothing. The plans' own foundation layout — 11" cantilevers, 61" spans, middle support dead center — was drawn with that 30% still in hand.

The lever we have: cantilever length

Cantilever and span trade one for one — every inch a support moves outward is an inch off the span between supports. Which end of that trade you want depends on which one is running hot:

• JCS is comfortable everywhere, so they sit at the conservative end. Short cantilevers keep tip deflection down at the corners, where sag would show and where the doors hang. It costs them nothing.
• We are span-limited, so lengthening the cantilever is the free lever for shortening the span. Going from 11" to 18" buys 7" off each span, and the cantilever end has capacity to absorb it.

Hence 18 / 55 / 55 / 16 instead of 11 / 61 / 61 / 11. Same 144" total; supports pulled inward; load shifted toward the end that had capacity to spare.

Why the two cantilevers aren't equal

Two legitimate reasons and no arbitrary ones:

1. The conduit. The as-built stub's north edge sits 9.19" from the skid's south end, so the south block can't come as far south as the north block can go north. There is no such obstruction at the north end.
2. The door. The gable walls load the cantilever tips, and they are not the same weight. The south wall has a 62½" × 73" hole in it filled with a plank door — roughly 32 of its 77 sq ft replaced by something about half the weight of shingled, sheathed wall. South gable ≈ 474 lb, north gable ≈ 617 lb. Less tip load earns a longer cantilever at the same stress, so the asymmetry points the right way.

Plan view — as-built foundation

Looking down. North up, east right (fence sides). Drawn to scale.

Note the middle blocks sit on the estimated root line. That is the one unresolved conflict — see Open items.

Edge-to-edge layout chart

Same plan, dimensioned the way you'll actually tape it — outside edge to outside edge in both directions. No center lines: every number is between two things you can hook a tape on. Drawn to scale at the actual block size.

Targets both ways — use the column that matches your block

Cross-pad dimensions (E–W)

Everything measured across the building. One table, five reference frames.

Self-checks that catch a wrong assumption

Along-skid dimensions (N–S)

Everything measured along a skid. Both skids identical, so every station gives a cross-pad pair to check.

Spans, cantilevers and clearances

Cross-section & vertical stack

Looking north. Shows why the rims float and where every height comes from.

The math

Every number above, derived. Estimates with stated assumptions — not a stamped calculation.

1 · Allowable bending stress

2 · Load takeoff

The critical distinction: bearing-wall and roof load reaches the skid as a distributed line load (via the rims and joist ends), but gable-wall load lands as a point load at the skid tips — the end joists cross the skids within the first 1½" of their length.

3 · Moments and stresses

Cantilever: M = P·L + wL²/2 · Interior support of a continuous beam: M ≈ wL²/10 · Stress: f = M / S

Allowables differ by species and service condition: 892 psi for our wet-service PT SYP, ~850 psi for JCS's dry hem-fir. Comparing the hemlock row against our PT number would be the wrong comparison — each skid is measured against its own.

The adopted layout sits within 4% of the pure optimum. That 4% was spent buying an extra inch of conduit clearance at the front block — a worthwhile trade for a dimension that has to be worked around with a tamper.

4 · Secondary checks — all pass with room

Bearing is why the blocks are turned 16"-across, and it is not the reason you'd guess. At 32 psi against 565 allowable — 6% of capacity — doubling the contact area by turning the blocks buys nothing. What 16"-across actually buys is ±5" of skid placement tolerance instead of ±1¼", and these skids sit on slick termite shields, get nudged during racking, and are positioned by string rather than a jig.

Field checks, in order

Open items

Phase 2 cut list — as approved Aug 5, 2026

Historical record, migrated Aug 14, 2026 from the retired Order_Plan_and_Build_Sequence. The deck is built; this is kept because it documents what was cut and why.

Ply layout: three columns across the 120" — 48" / 48" / 24" — with the N–S seams staggered (96+48 in one column, 48+96 in the next). The full 48×96 sheet goes at the SE corner so the conduit hole is measured off that sheet's own factory corner.

As-built & field notes — Phase 2, August 2026

Everything above is the design. This is what the deck actually taught us while it was going together on Aug 12–13, 2026 — corrections to numbers printed above, techniques that worked, and two things in this document that were simply wrong.

Scope note. This document covers base and deck only. Wall framing, roof framing and finishes get their own parallel documents rather than being appended here — the point of the library is that each phase is findable on its own.

Measure from the frame, not across open air

Skid-to-skid across open ground is the measurement that fought us all day: nothing to hook a tape on, nothing square to reference, and every reading disagreeing with the last. Once the deck frame is square, it is the reference. Measure down to the skids from it.

The deck is 120" wide and the skids sit at 99½" outside-to-outside, so:

The full chain across the deck must sum to 117":

8¾ + 5½ + 88½ + 5½ + 8¾ = 117"

The bottom edge of the rim sits level with the top of the skid, because the joists bear on the skids. So the tape runs horizontally with nothing to plumb down and no square needed.

Revised build sequence — supersedes the order printed above

Racking a frame that's out of square

Diagonals of 188" and 187" are 1" out. The correction is ½", not 1" — moving one corner lengthens one diagonal and shortens the other simultaneously.

Toenailing joists to skids — the gun mostly lost

Root causes, in the order they mattered:

Order of operations

1. Four corners first — both end joists × both skids. This pins the frame's outline, and from that moment no single nail can move the assembly.
2. Then center-out, halving the gaps: joists 5 and 6, then 3 and 8, then fill in 2, 4, 7, 9. Marching sequentially from one end lets each small push accumulate, and the tenth joist ends up visibly off.
3. Both skids on a joist before moving on. A joist nailed at one end only can pivot.
4. One nail from each side, opposing. Two from the same side walk the joist off its mark. Alternate which side you start on so any stance bias cancels across the deck.

2 per crossing · 20 crossings · 40 nails total. Measure the diagonals once after the four corners are in; if they still match, the geometry is frozen and the remaining 32 can go in without stopping.

Blocking — the two-pass method

The blocking rows back the plywood's long seams, so they cannot be staggered — every block in a row lands on the same line. Which means you cannot end-nail both sides of a joist at the same station: the block you just installed is in the way. Solve it with sequence instead.

5 end-nailed and 4 toenailed per row · 3 per end either way · ~108 nails for both rows. On a 5½" block: roughly 1½" and 4" up on the first face, ~2¾" on the second. Add a fourth if a block still feels lively.

Marking — one chalk line per row

1. Mark 23¼" on the north and south end joists, measured from the outside face of the east rim. The line marks the block's east face. (As-built Aug 2026: ply columns run 24 / 48 / 48 from the east, so the ripped 24" strip column lands on the east and the conduit hole falls in the SE strip piece — a 24"-wide piece drops over a stub-up far more easily than a full 48×96 sheet. The layout is rotationally symmetric across the diagonal, so either side works; the east was chosen in the field. Measure from the east rim only — two reference edges is how these get transposed.)
2. Snap. The line crosses all ten joist tops in one shot, and is straight by definition.
3. Put an X on the west side of the line on every joist — the block always sits west of its line. Without it you will set half the row on the wrong side and lose the seam bearing.
4. Repeat at 71¼" for the second row. Blocks occupy 23¼–24¾" and 71¼–72¾", centering the seams at 24" and 72".

Joist-to-rim gaps — leave them

Gaps of 1/32"–1/16" turned up at several joist ends, a couple of them consistent full depth.

The right method: clamp it, or run a temporary screw to draw it in, drive the nails, then back the screw out. Screw as clamp, nails as fastener.

The check: pull a string along the outside face of each rim, end to end. Fix only where it wanders more than ⅛" off the string, and only toward the string. Everywhere else, nail and move on.

Shims — as-built

Product: composite horseshoe shims, 1⅞" × 5⅛", ~$1.48 per 16-pack. Bought 5 packs / 80 shims for about $6.45. Parallel-sided, not tapered — no wedge tip concentrating load.

The load math

Contact patch is the skid's 5½" width × the block's 7⅝" along-skid dimension = 41.9 sq in. Three shims side by side — 5⅛" across the skid width, 1⅞" each along it — give ~29 sq in, about 70% coverage.

~1,400 lb ÷ 29 sq in ≈ 49 psi, against PT SYP's 565 psi allowable compression perpendicular to grain — 9% utilised.

The skid is the weak link, not the shim. Nothing here is close to troubled.

Bowed joists become blocking

Boards left on gravel a few days develop bow — the gravel side stays damp, the sun dries the top, and the board bends toward the dry face.

Yield

117 ÷ 14⅝" (14½" plus kerf) = exactly 8.0 — too tight to bank on once you trim a bad end. Budget 7 usable blocks per board. Against 18 blocks / 258 linear inches needed: