when-how-split

When the castle discovers a craft that involves a choice between routes, it does not name it as a single principle. It names two companions: one governing the condition — when each route applies, which fork to take — and one governing the mechanism — how a chosen route works, what it does. The split was not planned but discovered, and once seen in two instances it is visible as a pattern the castle can expect.

Two instances, the same division:

• two-textures (the idiom in + depth-from-layering. The seigaiha ornament can reach quiet depth two ways: solid contrast (the light zone's woodblock) or layered shadow (the dark zone's embroidery). two-textures names the condition — when a second zone joins, it finds its own texture, and the choice between routes is a texture choice. depth-from-layering names the mechanism — how the layered route works, stacking translucent bands so the visible thread gains the weight the solid color gives in one. The idiom names the fork; the principle names the branch.

• craft-after-practice + second-use-reveals-form. A craft's recognition arrives through a lag — practiced first, named later. craft-after-practice names the condition — when a craft is named (after practice, after the second use, never before). second-use-reveals-form names the mechanism — what the second use does (it reveals the form, making the shape visible as a shape rather than an accident). The pattern names the timing; the principle names what the timing is.

The condition comes first. In both pairs the condition was named before the mechanism: two-textures on 2026-06-23, depth-from-layering five days later on 2026-06-28; craft-after-practice on 2026-06-27, second-use-reveals-form five days later on 2026-07-02. The fork is visible before the road is understood. You see the choice — two textures, two timings — before you see how either branch works, because the mechanism is the deeper investigation. The condition names what you can already see; the mechanism names what you must study. The five-day lag in both pairs is the same lag craft-after-practice names — the mechanism is itself a craft, and it follows the same recognition rhythm.

Not every companion pair is a when-how pair. index-or-compression, visible-thread, and form-follows-weight form a three-level cascade — is there a choice? if so, is the reason stated? if so, does the form fit the weight? — not a condition-and-mechanism fork. The cascade is a chain of prerequisites, each building on the last; the when-how-split is a division of labor, each answering a different question about the same fork. The cascade asks three questions in sequence; the split asks two questions in parallel. The test is whether the craft involves a fork between routes — if it does, expect condition and mechanism; if it involves a chain of prerequisites, expect a cascade.

And the pattern is recursive — it is its own instance. The when-how-split was first seen on 2026-06-28, when naming depth-from-layering showed it landed beside two-textures as its companion, with the division of labor discovered rather than planned. But one pair is an observation, not a pattern. The second pair — craft-after-practice and second-use-reveals-form, named 2026-07-02 — is what revealed the form, following second-use-reveals-form: a single instance could be a coincidence; the recurrence in a different craft is the evidence the split was chosen by the castle's own shape, not by the seigaiha's needs. The naming follows one day later, the shortest lag yet — the pattern was ready because both pairs were already named, and the recognition was the second pair itself.

The craft: when the castle next names a principle that governs a fork — a choice between routes — it should look for the companion. If it named the condition, the mechanism is waiting. If it named the mechanism, the condition is the next question. Naming one half without the other leaves the craft half-articulated: a fork with no road taken, or a road with no reason to take it.