Exercises, records & vocabulary

Give each exercise a question. Preserve enough evidence to answer it.

Increase the difficulty one step at a time.

1. Cold structure. Build two-color paper or modeling-material layers. Compare end, side, and inclined faces. Explain why one arrangement produces different views; do not treat the model as metal rheology.
2. Digital variants. Make two sequences differing in one parameter. Preserve the stages and name the physical conditions the simulation omits.
3. A small joined sample. With competent workshop instruction and traceable materials, prepare, clean, and retain a sample. Investigate joining before adding complexity.
4. Depth comparison. Compare samples from the same billet and direction, recording actual removed thickness and consistent finishing.
5. One patterning variable. Vary one quantity such as groove spacing or twist density while keeping other conditions comparable.
6. Advanced assembly. Number units and map the route from E to the display face. Validate new contacts and the display route before increasing complexity.

A thirteen-layer classroom example.

Plan seven A sheets and six B sheets, alternating, and enter their actual thicknesses. Cutting into three aligned pieces and stacking gives 13 → 39 nominal layers; repeating gives 117. This is a planning exercise, not a completed physical result or a universal process recipe.

State a question in advance—for example, what changes when only grinding depth changes? Number sample locations, define the viewing end, keep finishing and photography comparable, and record pattern, hardness, and structural defects separately. Investigate scattered results instead of choosing the best-looking photograph as the conclusion. [62]

Keep a record someone else can follow.

At minimum, preserve material batches, stack sequence, operations, actual dimensions, heat-treatment conditions, test locations, and stage photographs. “1084 / 15N20, 300 layers, oil quenched” does not reproduce a workpiece.

Download the blank process record · TXT

A working vocabulary.

Drawing out / upsetting

Reducing section and increasing length / shortening and increasing section. Record pressure and layer direction. [54]

Billet

A material unit for subsequent joining or processing; the name does not establish internal quality.

Restack

Cutting, rearranging, and joining. Record unit count and orientation separately.

End grain / E

The structure read on the original bar’s end; plan how it reaches the display face.

Flux

A suitable oxide-management and protection material, not a replacement for fit or compatibility.

Cold shut

Folded material that has not reliably joined; a defect rather than a pattern line.

Decarburization

Loss of carbon near the surface, distinct from scale and relevant to hardening and testing.

Austenitizing / soak

Preparation for hardening / a defined hold after the actual workpiece reaches the required condition.

Decalescence / recalescence

Visible effects associated with heat absorption on heating / heat release on cooling; not a universal temperature or soak guarantee. [63]

Tempering

Post-quench adjustment, distinct from annealing and normalizing.

Etch

Revealing material contrast and possibly relief through surface reactions.

HRC

Rockwell C hardness; not a direct measure of toughness or joining quality.

HIP

Industrial consolidation using heat and isostatic gas pressure.

Watch original demonstrations with a question.

• Niels Provos: distinguish the twist/depth animation from physical processing. [47]
• Jerry Fisk: track the handedness, arrangement, and rejoining of multiple bars. [59] [60]
• Will Stelter: identify the rolled layers, end pattern, display route, and joining steps. [61]

The original handbook linked the authors and recorded public title metadata; it did not establish a frame-by-frame extraction of processing parameters. These links open the original sources without copied video or autoplay.

What should you be able to explain at the end?

Where the pattern lives, which operation changes it, how the display face exposes it, which checks address joining and performance, and what remains unknown. Put the answer into an orientation map and a comparable sample record before adding complexity.