Description
A knot fails in one of two ways. It breaks, which you find out about immediately and can do nothing about; or it slips, which you find out about slowly and usually blame on something else. Most connections lost in a season are the second kind, and almost all of them were visible before they went.
The Knot Fault Atlas is a reading instrument for that second case. Twelve connections that actually carry load — improved clinch, palomar, uni, snell, double uni, blood knot, Alberto, FG, surgeon’s, non-slip loop, bimini twist and loop to loop — each drawn three times from a single geometry with exactly one thing changed between the drawings: a turn removed, a collar moved to one strand, a barrel left standing off the eye. If the right drawing and the wrong drawing differed in six ways you would have to hunt for the one that matters. Changed once, you cannot miss it.
The bench read. Pick the knot in front of you and answer only what you can actually see. The atlas returns one of four verdicts: fish it, holding, retie now — or CANNOT BE READ FROM HERE, which is what it says when the feature that would decide the question is buried inside the knot, as it always is on an FG and a welded loop. It also escalates: a holding knot is fine on a bluegill afternoon and is not fine when you have driven three hours, and it will say so if you tell it what you are hoping for.
What you get
- A self-contained offline browser edition — no account, no server, nothing uploaded, works with the signal off.
- A 19-page Letter print edition, legible in grayscale, with a blank bench card for the connections in the rig you are about to fish.
- A local working record on your own device, with JSON export and import so your notes stay yours.
Poor fit
If you want a ranking of the strongest knots, this is the wrong document. It does not rank, and it will not repeat bench-test percentages as though they survived contact with a fish. It is also no use without a knot actually in front of you.
Your purchase includes the browser edition and the print edition, and any corrected release of this edition at no further cost.
You may read and print this on your own devices as often as you like, use its methods and working pages in your own work, and — if you work with clients — use it while serving them and hand a client a completed output you produced with it.
You may not redistribute the publication itself, the blank working masters or the print file; resell or republish it; put the complete work in a shared library or community drive; or share your purchase credential in place of a purchase. It may not be used for model-training ingestion.
Refunds: fourteen days, no reason required. A broken, missing or materially misdescribed file is corrected or refunded without relying on that cutoff.










