There is an old Windsor in a lot of shops, the kind that came off a wagon a hundred years ago and still sits four-square on the floor. Pull it apart and you find cut nails in the rungs. Now look at a reproduction built two years ago with structural screws in the same joints. It already rocks. Same chair, same loads, and the older one with the cheaper fastener is the one that held.
That looks backward. It is not. It is the most counterintuitive thing about fasteners, and once you see it you stop reaching for the screw out of habit.
Start with what a fastener actually does, because there are two jobs and most of us only compare one. The first is pull-out. Load the fastener straight along its length, try to draw it out of the wood, and the screw wins every time. Threads beat a smooth shank, that is not in dispute. The second job is lateral shear. Load the fastener sideways, across its length, the way a chair rung gets loaded every time someone drops into the seat and leans back. That is a different test, and the screw does not automatically win it.
Here is the part nobody tells you. When a fastener is loaded sideways, what matters is not how much load it takes at the peak. It is what happens at the peak. Some joints fail brittle. They hold and hold and then let go all at once, no warning, no give. Other joints fail ductile. The fastener bends, the joint deforms, it absorbs the shock and keeps holding well past the point where the brittle one already quit. A chair sees that load over and over for years. In a connection like that, the joint that bends and holds beats the joint that is stiffer but snaps.
The nail bends and holds. A big structural screw in the same spot often does not. The screw is thicker and the steel is harder, so under a sideways load it tends to crush the wood around it and fail fairly abruptly. The nail is thinner and softer, so it forms a hinge and keeps taking load as it deforms. The very thing that makes the nail weaker in pull-out, less steel, less stiffness, is what lets it bend instead of break when the load comes from the side. It is not that the nail is stronger. It is that it gives. Those are different things, and for a chair you want the one that gives.
This is why the traditional choice was not ignorance. Windsor and stick chairs got nailed because the makers had watched a hundred chairs come back, and they knew which joints stayed tight. They could not have drawn you the engineering. They did not need to. They had the result, and the result was a fastener that deformed instead of fracturing in exactly the joints that flex.
So the move at the bench is to sort the joint before you sort the fastener. Is this connection loaded sideways, and will it see repeated load? A chair rung, a seat rail on a chair that rocks, a stretcher on a stool, a tool tote that gets swung and set down all day. Those are lateral, cyclic joints, and they belong in the column where a nail or a slender fastener that bends will outlast a fat screw that snaps. Reach for the screw there because it feels like the upgrade and you may be building in the failure.
Hold the line on the other half, though, because this is narrow on purpose. The screw still wins when the load is pull-out, when something is being drawn straight off a face. A cleat carrying a shelf, a back panel pulled away from a carcase, anything in withdrawal. There the threads earn their keep and a nail would be the wrong call. The fastener decision splits clean by direction of load. Sideways and repeated, the slender fastener that bends. Straight out, the screw.
None of this means you go back to nailing everything, or that a screw in a chair rung fails tomorrow. Plenty hold. But when you are standing over a dynamic, sideways joint and the instinct says screws are the stronger choice, that instinct is answering the pull-out question for a joint that was never going to fail in pull-out. The old-timers were answering the right one.