The groove matters more than the ring
Gland design, and the dimension that decides whether a seal works.
The groove matters more than the ring because the groove, properly called the gland once its full cross-section is considered, decides how much the ring gets squeezed once the joint is assembled. A first-rate ring in a badly sized groove is squeezed the wrong amount however good it is, while an ordinary ring in a correctly sized groove seals reliably because the groove around it was designed with the same care as the ring inside it.
Depth sets the squeeze, width sets the freedom
An O-ring's own quality (its material, roundness and surface finish) sets an upper limit on how well it could seal under ideal conditions, and the groove decides whether that potential is realised. The groove's depth relative to the ring's cross-section directly sets the squeeze described in the previous article. Its width decides how much room the ring has to move, stretch or roll as pressure arrives from one side. A ring too free to roll in an overly wide groove can twist under pressure and present a less effective part of its cross-section to the sealing surfaces, a failure that owes everything to the space it was given and nothing to how well it was made.
A groove cut too deep starves the ring of the compression it needs, and one cut too shallow over-compresses it. Both failures trace back to a single dimension on a drawing that nobody looking at a finished joint would think to check, since the groove disappears from view the moment the ring is installed.
A picture that will not hang flush
A picture hung on a mount with the wrong depth of recess either sits proud of the wall or sits too deep and strains the hardware, and the picture and frame can both be flawless while the recess is the reason the assembly sits wrong. The groove around an O-ring plays the recess's role, an unglamorous, mostly invisible piece of geometry that decides whether whatever is fitted into it behaves as its own quality would suggest. Nobody blames the picture for a badly cut recess, and a seal failure that traces back to the groove is no fault of the ring either.
Swap in a better picture or a more expensive frame and the recess still refuses to sit flush. Fitting a premium O-ring into a badly sized groove fails for the same reason, because the improvement has gone to the wrong half of a two-part system.
A tenth of a millimetre on the drawing
Gland design standards specify groove depth, width and surface finish to tight tolerances, and simple arithmetic shows why. On a ring with a 2 mm cross-section, a groove cut just 0.1 mm too deep removes five percentage points of squeeze, a fifth of a typical 25 percent target, from an error about the thickness of a sheet of paper. That sensitivity is on a par with the roundness tolerances the earlier set on pressure housings described for a different structural reason. A housing built to a demanding depth rating gains nothing from that rating if the one small groove holding its main seal was never held to the same standard, so a designer confident about the depth rating has a concrete question to ask next: were the groove tolerances on the drawing held as tightly as the shell's wall thickness, or left looser out of habit?
Gland dimensions therefore belong on a drawing with the same rigour and checked tolerance as any other critical dimension, and a groove left as a rough standard slot for the machinist to judge is a seal left to chance. A design review that scrutinises the ring's material and size while glossing over its groove has checked only half of what decides whether the seal works. A supplier's catalogue of standard ring sizes is useful only when paired with an equally deliberate groove drawing, and treating the ring size as the whole specification hands away the control the standard was meant to provide.
Where groove tolerance can be looser
Groove tolerance matters most for static seals under significant pressure, where a small squeeze error compounds into a real risk of leaking. It matters somewhat less for lightly loaded seals, or for dynamic seals designed with deliberately looser tolerances to allow for movement, where the ring's resilience absorbs small variations without much consequence. Even there, a seal remains a system of at least two parts, and treating either in isolation misses the kind of interaction that decided the fate of the cable entry this set keeps returning to. The next article turns to a groove that was cut correctly and then over-tightened anyway, a different failure from the same root cause of mistaking one dimension of a seal for the whole of it.