A TIG torch walking a longitudinal seam on a cylindrical shell, filler wire feeding in from the right, and behind the arc a line of overlapping crescents that give the technique its nickname. Under the shell, clamped to the bench, are shaped copper blocks — chill bars — and along the joint a few small tacks hold the two edges in line before the run gets there.
The dimes are what the camera was pointed at. The copper is what made them possible.
A stack of dimes is not a surface finish applied at the end. It is the frozen record of the weld pool advancing one filler addition at a time.
The arc establishes a pool. The welder dips filler into the leading edge of that pool. The pool moves forward, the trailing edge freezes, and the next dip lands partly on fresh metal and partly on the previous crescent. Repeat at a steady interval and the crescents overlap into the pattern. Change the interval and the pattern changes with it.
So the spacing is not decoration, it is arithmetic: travel speed against dip frequency. Move faster with the same rhythm and the dimes stretch apart. Dip faster at the same travel speed and they crowd and stack higher. Hold both steady and the bead looks like it was drawn with a ruler, which is exactly what people photograph.
What the arithmetic does not tell you is heat. Two beads can have identical spacing and completely different amounts of energy put into the plate, because current and travel speed can trade against each other. The pattern records rhythm. It does not record amperage.
The seam in this frame is longitudinal, on a cylinder that can be rotated or walked along. That geometry is generous. The welder can set a posture, keep the torch angle fixed and let the work move past the arc at a constant rate. Rhythm becomes a physical habit rather than a continuous correction.
Put the same welder on the outside of a tight bend, or on a short joint where the torch has to start, stop and change angle within a few centimetres, and the dimes will not be identical. That is not a decline in skill. It is the geometry charging rent. A shop that shows only straight seam beads is showing you its easiest work.
This is also why fixtures matter more than they look. A positioner that turns the part at a constant speed removes one variable from a job that has four or five. Anything that reduces the number of things the hand has to do at once improves the rhythm for free.
Those copper blocks are not there to look purposeful. Copper conducts heat away far faster than steel does, and clamped against a thin wall it acts as a heat sink on both sides of the joint.
Without it, welding a longitudinal seam on a thin shell puts heat into a wall that has nowhere to send it. The wall around the bead expands, the bead contracts as it freezes, and the shell dishes. On a cylinder the distortion shows up as an oval section, or as a seam that pulls the ends out of line so the next part no longer fits.
With copper on both sides, a share of the heat leaves through the blocks instead of through the shell. The welder can then use enough current to achieve fusion without cooking the surrounding wall, and the shell stays round.
Two things worth being precise about. First, copper here is a thermal tool, not a shielding tool: it stops heat, it does not stop oxygen. On metals that need the back of the weld protected, gas still has to be arranged separately. Second, the copper has to touch. A chill bar sitting a millimetre off the surface is a decoration, and the difference is invisible in a photograph.
Here is the part the industry does not like saying out loud. A perfect stack of dimes is not evidence of a sound weld.
A weld can have textbook ripples and still fail on any of these: lack of fusion at the toe, where the bead sits on the plate without having melted into it; insufficient penetration, where the root never closed; undercut along one edge, hidden in the shadow of the ripple; or excessive reinforcement, a bead piled high enough that the ripple pattern is the only thing anyone looks at.
All four are cosmetic-pass, structural-fail. And all four can be produced by someone who has practised the appearance and not the process — which is why a portfolio of bead photographs is weak evidence compared with one bend test.
The pattern is also easy to over-produce. Running slower and dipping more heavily gives taller, more dramatic dimes, and it puts more heat and more filler into the joint than the section needs. The most photographable bead is often the one that was worked hardest, not the one that was made best.
None of that makes the appearance worthless. It proves something narrower and still useful: that the welder held a rhythm over that length of joint.
Holding rhythm means the hand, the wire feed and the travel all stayed coordinated. That is a real skill and it correlates with consistency — a joint welded in a steady rhythm is more likely to have uniform penetration than one welded in a stop-start panic. Irregular dimes are a warning sign, not a verdict: they usually mean the welder fought the position, lost the pool or stopped to let heat out.
So read the pattern as a process indicator. Even spacing says the operator was in control. Width that stays constant says the gap and the heat stayed constant. Then go and find the evidence that actually matters.
Pulsed TIG can lay down evenly spaced ripples with no dipping at all: the current pulses, the pool advances, and the solidified surface takes a regular scalloped shape. The result photographs impressively and is genuinely consistent, because a machine does not get tired at the end of a seam.
It is a different thing from a hand-stacked bead, and telling them apart is not hard once you know what to look for. A hand stack varies slightly in crescent width, height and overlap, because a human is maintaining rhythm against real feedback. A machine stack is closer to identical, and a close look often shows the pulses sitting on top of a smoother underlying surface rather than each crescent having been individually fed.
Neither is automatically better. A consistent pulsed weld is excellent for repeatability on a production part. A hand-stacked bead tells you the person was reading the pool. What matters is that a shop can tell you which it is doing and why, rather than presenting the pattern as proof of craft.
Judgement of a weld belongs on the far side of the joint, and on its edges.
Look at the toe lines: they should meet the plate smoothly, with a wet angle rather than a crease. Look for undercut, which shows as a groove along either edge of the bead. Look at the width: if it tapers, the heat or the gap changed. If the back of the joint is accessible, look for a consistent root — full penetration shows there, not on the face.
Then get a coupon. A section cut across the weld and etched, or a bend test on a sample from the same setup, answers in ten minutes what a photograph cannot answer at all. Any fabricator who will not cut up a test piece for a new joint type is asking you to take the bead on faith.
Put a straight edge along the seam — the shell should not have dished, and a chill bar that was actually touching usually shows in that result. Look down the bead at a low angle so you see the profile rather than the pattern; a bead should be slightly convex and even, not a row of peaks. Check both toes for undercut. Look at where the seam ends: starts and stops are where lack of fusion lives, and a shop that dresses them neatly is usually a shop that watched them.
And ask the question that separates a process from a performance: what was the current set to, and by what rule. Someone welding to a procedure can answer. Someone chasing a photograph of dimes usually cannot.
