Tips for your first trench knife build that I learned the hard way

Nov 11, 2026 Last reply: 8 hours ago 4 Replies
Jump to replies (4)

Been seeing a few trench knife builds pop up lately so figured I'd throw out some stuff I wish someone had told me before my first one. The big one is the knuckle bow. Don't try to bend heavy stock cold unless you enjoy snapped tangs. I heat the tang section to a nice orange and bend it around a form pipe slowly, checking it against my paper template every few passes. Go slow. Once it's bent too far there's no unbending without trashing the profile. Layout matters more than people think. Before you cut anything, trace your actual hand on paper holding a mock handle blank and mark where each finger naturally sits. My first attempt had the notches so close togehter my pinky just refused to cooperate. I now space the finger ridges by wrapping masking tape on a dowel and gripping it untill the placement feels right, then measuring off that. Bog oak is gorgeous but it can move on you. Even wood that's been buried for centuries needs sealing, expecially around the pins where moisture sneaks in. I drill my pin holes a touch oversize and soak the scales in thin cyano before final assembly, then use a mosaic pin as the centrepiece since the dark wood shows it off well.



Grind the notches before final glue-up. Way easier to file and sand those ridges with the whole thing still in pieces.


I spent a few weeks working on a similar piece and realized the advice about the knuckle curve is helpful, but I found the bending process needs a little more stability than I anticipated. Instead of bending it around the pipe straight away, I started clamping the tang section to a piece of scrap metal first. That gave me better control over the initial angle before running it over the forming tool. It saved me from having to undo a major set of bends later.

When dealing with the ancient wood, the issue wasn't just moisture, but the uneven density of the burl itself. Even after sealing the surface, the pins kept shifting slightly. I ended up using a specific type of brass shim inside the handle area to lock the scale into the handle blank before I glued it down completley. It made a big difference in how solid the final result felt. I used a minimal amount of epoxy around the rivets instead of cyano, just for that extra set of anchors. My biggest surprise was the sanding. I tried finishing the ridges while they were still separate pieces, as suggested, but it felt too delicate. I waited until the handle was fully assembled and then used a high grit belt to take down the edges. It looked cleaner and was much more effective than trying to finesse the small grooves before the adhesive was applied.

On the bending part, I've started normalizing the tang a couple of times before the final bend pass, mostly because I kept getting a slight springback that put my curve just shy of the template. Working the steel at a duller heat in short passes seemed to give a truer set than one confident bend, though I've had a friend argue the opposite so who knows. Also seconding the hand tracing idea with one addition: trace it while gripping the actual thickness of stock, not a pencil blank, since even a few millimetres of extra bulk changes where the fingers land more than I expected.

The part about gripping a taped dowel to find the finger spots is such a simple idea. I tried it last night with an old broom handle and some tape and it allready showed me my notches were too far apart, which explains why my hand slides forward when I hold my practice piece. One thing I keep wondering about though, that fancy pin with the pattern in the middle, does the dark old wood actually hide it or does it really pop? Also how do you keep the wood from splintering when you drill into it, mine keeps chipping at the hole edges.

Quick update — I gave the idea of normalizing the tang a try before the final curve. Doing a few passes realy helped minimize that springback I was seeing. It seems working the steel at a slightly lower heat in smaller segments is giving a much truer set than I expected.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required