Design record · superseded · not part of the site

This line was not taken. It is kept because it is the evidence behind the decision that was: a hull generated from parameters will not reach the density of a real measured survey, and the four sketches below are how far it got before that became clear. Anyone asking later why the site quotes someone else's drawing should read this record first — the alternative was built, not assumed.

What shipped instead is a public-domain plate from the Historic American Engineering Record — HAER MA-168, Schooner Ernestina — cut into two views: the deck plan standing in the right column of /architecture, the inboard profile closing /practice. See /design/ship-plate.html for the treatment and /design/ship-elements.html for the arrangement.

A ship drawing for the architecture section

Sketches of one hull, drawn several ways, to settle which drawing the page should carry and how much sheet it needs to stay legible. Sketch 04 was the furthest this line got — the layout of the two reference sheets Proteus supplied, a working fisherman's inboard profile over her deck plan. Sketches 01–03 are the earlier attempt, a lines plan, kept because the hull beneath every one of these is the same hull. None of it is installed on the site, and none of it now will be.

The hull is ours. No scan, no third-party asset, no licence to carry alongside it. Every curve is sampled from a closed-form parametric model — deck half-breadth, sheer, keel with drag, and a section fullness that runs from a firm bilge amidships to fine ends — so the three views agree with one another the way a real lines plan must. They are one surface projected three ways, not three drawings that resemble each other. Move a station in the model and all three views move together.

It is not a named vessel and does not claim to be. Anyone who knows hulls should find nothing wrong with it; nobody should be able to say which ship it is.

Sketch 04 · Inboard profile over deck plan

The survey pair — two views, one scaleMasts drawn in both
FORECASTLE FISH HOLD CABIN BOWSPRIT STEM RABBET KEEL FALSE KEEL WORM SHOE KEELSON STERNPOST COUNTER FRAMES DECK BEAMS WINDLASS FORE HATCH MAIN HATCH CABIN TRUNK WHEEL MAST PARTNERS 0 10 20 30 40 50 60 70 80 0 4 8 12 16 20 24 FEET METRES WORKING SCHOONER INBOARD PROFILE AND DECK PLAN 72 FT BETWEEN PERPENDICULARS NOT A NAMED VESSEL · LATENTWINDS

The layout both reference sheets use, and the reason they read as surveys rather than as pictures: the same objects seen twice. Every hatch in the deck plan stands at the station of the hatch in the profile above it, because both views are generated from one list of fittings. Declaring them twice would let the two views drift, and a survey whose views disagree is not a survey.

The masts carry the only colour, and they earn it: they are the one element that appears in both views, so an eye following a mast down from the profile lands on its partner in the plan. Theresa picks hers out in blue and red for that reason, not for decoration.

The rig is cut off, as it is on both references. A rigged sail plan is roughly 1.55:1 and a different drawing; this pair is 1.9:1, and one view alone is 4.4:1 — which is the ratio that fits a right-hand column.

And here is the cost. Fifteen labels at 10.5px need close to the sheet's drawn width, so this scrolls sideways on anything narrower than a large desktop. Annotation and a narrow column are mutually exclusive: the rail version has to drop the labels, or the plate has to run full width and the labels stay. That is a decision, not a defect, and it is not one this record can take.

The grid is ten feet, not ten millimetres: it is the vessel's own measure, and the scale bar beneath reads in feet and metres from the same one assigned dimension — 72 ft between perpendiculars. Nothing else in the model has units.

Sketch 01 · Sheer plan alone

Sheer plan — the hull in profileTen stations · five waterlines · three buttocks
0 1 2 3 4 5 6 7 8 9 10 WL 1 WL 2 WL 3 LWL WL 5

The cheapest option and the one most likely to read as decoration. A profile alone is a silhouette: it says ship and stops. The buttock lines are the only thing in it doing engineering work, and they are easy to mistake for styling.

Sketch 02 · Sheer and half-breadth

Two views to one scaleStation verticals shared between them
0 1 2 3 4 5 6 7 8 9 10 WL 1 WL 2 WL 3 LWL WL 5 BUT A BUT B BUT C

The second view is what turns a silhouette into a drawing. The same ten verticals cross both views, so a reader can see that the two are measurements of one object rather than two pictures of it. This is the point at which the sheet starts to argue the section's case: the vessel is fully described before it exists.

Sketch 03 · The full lines plan

Sheer · half-breadth · bodyForward sections right, after sections left
0 1 2 3 4 5 6 7 8 9 10 WL 1 WL 2 WL 3 LWL WL 5 BUT A BUT B BUT C 0 1 2 3 4 5 6 7 8 9 10

The body plan is the view a shipwright actually lofts from, and it is the one that carries the site's own word: each curve is a station. Forward sections to the right of the centreline, after sections to the left, which is both the convention and the only way the fine bow sections and the full quarter sections stay legible.

It also costs the most sheet. At 60rem minimum width this scrolls sideways in its own box on anything narrower than a laptop, rather than shrinking its labels past reading.

What is still open

Where it goes. Not decided. The candidates are a frontispiece between the intro and the four cards, a band closing the page beneath them, or the right-hand third of wide screens, which that page currently runs empty.

Whether it is numbered. The E-numbers belong to destination pages — E.1 through E.4 are the four cards. An index-page drawing is not a fifth destination, so it either carries no number or needs a designation that says what it is. Minting E.0 would be this record inventing a series convention, which is not its to invent.

Whether the stations get named. They are numbered 0–10 here, as a lofting drawing numbers them. Naming them after system stations would tie the drawing to AM–D04's subject and put AstralMaris specifics on a general architecture page.