Every joint in fabricated metalwork sits somewhere between loose and tight, and where it sits decides how the weld behaves. A slip fit assembles by hand and leaves a gap. An interference fit needs force and leaves no gap at all. A weld gap is a deliberate clearance at the root, chosen so the arc can reach the bottom of the joint.
These are not three grades of the same thing. They are three different intentions, and mixing them up is one of the commonest reasons a joint that looked fine in tack welds fails a leak test or a bend test.
A slip fit is the workhorse of tube fabrication. One tube slides inside another, or into a socket, and the joint is located by the overlap rather than by contact at the end face. It self-aligns, it tolerates a small angular error, and it can be tacked in place and then adjusted before the final weld.
The cost is the gap at the root. A slip fit with too much clearance leaves a cavity that the arc cannot reach, and the result is a joint that is fused at the surface and open underneath. The fit has to be close enough that the gap is consistent all round, not just at the point where it was measured. The same requirement governs any fabricated assembly built from several joints, because one loose fit forces every later joint to compensate for it.
An interference fit is a deliberate pre-load. The parts are pressed together so that the interface carries load by friction and the joint has no gap to fill. In fabrication it is used for inserted components such as a boss in a plate, a spigot in a bore, or a stub in a tank wall.
The problem is that a pressed joint cannot be welded at the interface, only around it. The weld becomes a seal and a mechanical retainer rather than a fused connection, and if the fit relaxes when the part gets hot, the joint is relying on a seal that was never designed as one. Interference suits shrink-fitted parts and bearing seats; it is usually the wrong instinct for a welded tube joint.
The weld gap exists for the arc, not for the fitter. A root opening lets the arc reach the bottom of the joint so the two walls fuse to each other rather than to a bridge of filler sitting on top. On a butt joint the gap is chosen with the wall and the process, and it has to be consistent for the length of the joint.
Consistency is what a hand-fitted joint usually lacks. A gap that is correct at one point and closed at another means the welder is changing technique along the joint, and the section that was closed will be fused at the surface with no penetration behind it.
Heat closes gaps. As the weld pool forms, the material either side expands and then contracts on cooling, and a joint that was open at the start can be shut by the time the arc reaches the far end. That is why a joint welded from one end to the other often shows good penetration at the start and none at the finish.
The countermeasures are sequence and restraint: alternate sides, backstep, or tack more heavily than feels necessary and let the tacks hold the gap. On thin material the gap can also open rather than close, because the edges pull away as they heat, and the same sequence logic applies in reverse.
Both behaviours are covered in the stainless TIG welding notes, because a gap that moves is a procedure problem rather than a fitter problem.
A feeler gauge is the honest tool, and it should be used at several points around the joint rather than one. On a tube joint the gap at the top of the circumference is not the gap at the bottom, particularly if the cut is slightly out of square. Four checks at the quarter points will show whether the joint is consistent or merely close somewhere.
Where a gauge is not available, the shadow line test works: hold the joint against a light and look for the gap that shows. It is crude, it is fast, and it catches the gross error that a single measurement misses. What it will not do is tell you the gap is uniform, which is the property that actually matters.
A loose fit needs more filler, more filler needs more heat, and more heat means more shrinkage. So a joint that is generous to assemble is a joint that will pull harder as it cools. The fit and the distortion are not separate topics; the first causes the second.
The reverse is also true. A joint that is too tight cannot be fused at the root, so the welder turns up the current to force penetration, and the extra heat distorts the part around a joint that had no gap to start with. Both directions end in movement, which is why the target is a consistent gap rather than the tightest one achievable.
If every joint in a batch needs the same correction, the drawing is asking for something the process cannot deliver. A nominal dimension with no allowance for cut width, bend stretch or weld shrink will always produce a joint that has to be forced. That is a specification problem and it should be fixed at the source rather than at the bench.
A single bad joint in a batch is different. It points at one cut, one setup or one operator, and it is worth investigating rather than correcting. The distinction is easy to make if the fits are recorded, and impossible if they are only judged by feel.
Fit is a specification, so it belongs on the drawing with the joint it describes.
Reading a fit is mostly a matter of knowing which of the three you are looking at. A joint that assembles easily is not automatically good, and a joint that needs a tap is not automatically accurate. The stainless welding pages treat gap as the first variable in the procedure, the exhaust pipe range shows the fits used on production tube, and titanium fabricated components covers the tighter end of the range. Thin-wall fit-up is discussed under aluminium welding, and where a joint has to match an existing part the customization process begins with gauging the fit rather than the dimension.
