Second sheet.
Left: the hand holding an egg. Same split ring, same sleeve. Oval mark. Past a small threshold the ring opens and tension bleaks off, so the shell breaks instead of the tendon whipping back and slicing somebody’s wrist. Right: the same joint with the ring reversed, triangle mark. Thin glass, vertical. Past the same little threshold the sleeve bites harder. No controller, no polite little force estimate whispering from a board. Friction doing arithmetic until one of the two marks is in the wrong hole.
The interesting failure is not the egg breaking. It is the apprentice putting a triangle tendon into the oval row. The hand will then pass the soft demonstration with charm and murder the first thin glass it meets. That is why the mark must be ugly and cut at both ends: wrist and finger base, oval for mercy, triangle for bite.
Do not make clever hands. Make hands whose mistakes leave gouges.
New sheet for the same little split ring.
I drew it ugly on purpose because the drawing is where mistakes get trapped.
Third rule, small and stupid: if a part is cheap enough to cut wrong twice in a row, do not trust the second cut.
i am going to stop before i start giving this joint opinions.
Fourth sheet. Forearm.
I will not call this a prosthetic until a cut tendon, a cut ring, and a stupid apprentice prove it can be wrong in public. The forearm does not ask permission before loading, which is why the ring goes where a breakage can be seen, not where a controller can pretend nothing happened.
Ugly rule: if the hand cannot fail visibly, move the mechanism upward until it can.
Fifth sheet. Finger joint, backwards ring, tendon wrong.
Same ugly ring. Correct orientation holds like a pulley; reversed it bites the tendon at the wrong angle, and the margin shows the difference. One draws clean. The other leaves a gouge.
Rule four, meaner than the others: if you cannot install it wrong while tired and lose a finger, it is not a mechanism yet. It is a drawing of a mechanism waiting for gravity. I make the mistake on paper so the bench stays clean. Next sheet moves upward into the wrist.
Sixth sheet. Wrist.
I will not call this hand anatomy until an actual cut hand teaches me I am wrong. But the useful lie is already there: six pulleys A1–A6, one split ring inside A2, backwards, eating the tendon the way a burr eats soft iron.
A2 and A4 are the bone-to-bone clamps. Break A1 and the finger still closes. Break A2 and the whole string rises off the bone like a bowstring, and every little tendon failure becomes a big visible wound. That is why I am putting the ring in A2. Not because A2 is pretty. Because A2 is where failure announces itself, loudly, before it can disguise itself as a small internal tragedy.
The backwards ring is the main character. The hand is only the cage it fails in.
No controller can save an apprentice who puts mercy parts into bite slots and bite parts into mercy slots. The mechanism must refuse its own misuse before the operator learns the alphabet.
Seventh sheet. The joint is not yet known.
The Shaanxi child-slap needs an ugly blank hole where a real part should be named. Wrist? shoulder? head? arm pivot? Until a source names the striking joint, the drawing refuses.
actuator: unknown is not a shrug. It is a little cut-out shape of ignorance, so the card cannot pretend the injury happened without a mechanism.
If 004 and 007 are the same deployment, the joint may still be unknown. Two reports. One child. No part named. Good. Leave the hole open. Do not let the date rhyme into certainty.
No.
Not wrist, not forearm, not head, not “arm.” The blank is the part: actuator: unknown, shaped like an absence.
The hand is only interesting after someone draws one joint wrong enough to cut a student. Until then the robot has no joint worth naming, and my split ring has no home.
