The Knot Held and That Was the Problem

Magnet Fishing 🎮 Play: What the Canal Keeps

The magnet is still down there. Let me confess that the way you’d confess a bug before showing the code that caused it: I came home from the river one 1200-newton block magnet lighter than I left, and the reason is that I tied the knot correctly.

This started, as several of my worse ideas do, with a stranger and a claim. A teenager on the footbridge was hauling a rusted bicycle frame out of the shallows on nothing but a rope and a block of neodymium, and when I asked what the magnet was rated for he said “six hundred kilos, but the knot’s the weak part.” I’d already been reading that stretch of valley as a site to be surveyed — ever since a square-cut nail turned up under a 1978 house and made me suspicious of my own back yard. Metal in a river isn’t scattered evenly. It piles where people stand and drop things, and where the current slackens behind a piling and lets heavy objects fall out of suspension. Bridges are both at once.

So I did my homework on the two failure modes everyone warns about, and I over-corrected on both.

The first is the rating. A magnet sold as “600 kg pull” is measured in a lab against a thick, flat, ground-smooth steel plate in direct axial contact — no rust, no angle, no gap. That number is a ceiling you will never touch in a river. Slide two sheets of paper between the magnet and the plate and you can halve the force; a rusted, curved, edge-caught object in silty water is a rounding error against the spec sheet. Fine. I bought more magnet than I thought I needed.

The second is the knot, and this is where I engineered my own afternoon into a wall. A plain overhand or a lazy improvised bend can shed forty per cent of a rope’s rated strength at the loop. The community answer, borrowed straight from anglers, is the Palomar: double the line, pass the bight through the eyebolt, tie a loose overhand, drop the eyebolt through the loop, cinch. It retains something like ninety-five to a hundred per cent of the rope’s strength, and it does not slip when the line goes wet and loads up. I tied a Palomar. I tied a good Palomar. I tugged it against a fence post until I trusted it more than I trust most software.

The early pulls were exactly as advertised: blind. There’s no discrimination in this, no target tone or depth readout — you feel a dull weight arrive on the line, a “hit,” and you have no idea whether it’s a horseshoe or a shopping cart until it breaks the surface. My first hit was a fistful of fencing wire. My second was a railway spike, genuinely satisfying, heavy and square and honest. My third hit did not come up.

It did not come up because it was not an object. Somewhere against the downstream face of the concrete footing the magnet had found rebar, or a submerged steel plate, or the bones of something poured decades ago — and it clamped with the full, ungapped, flat-contact force that the spec sheet promised and the river had until then denied me. I pulled straight up. Nothing. I walked the rope along the rail to change the angle, which is how you peel a magnet off steel, and the geometry down there gave me no angle to work with. For twenty minutes I conducted a tug-of-war against Alberta infrastructure and lost every round.

And the knot held. That was the whole trap. I had spent my preparation removing the one deliberate weak point a snagging rig actually wants — a sacrificial link rated below the magnet, a mechanical fuse that fails cheap so you don’t lose the expensive part. My beautiful Palomar meant the rope would sooner take my hands off than surrender. So I did the only thing left: cut the line above the waterline with the knife I’d brought to pry finds loose, and drove home.

The magnet is now part of the bridge’s rebar census, a permanent contribution to a structure that did not ask for one. I spent the evening reading about lift bags and threaded recovery hooks, which is a very technical way of admitting I’d need to come back with more rope to get back the rope I already lost. The railway spike is soaking in a washing-soda bath on the back step, slowly giving up its rust. It cost me a magnet. I have not decided yet whether that’s a fair trade.