O-Ring Seal

Measure the O-ring's cross-section diameter (the thickness of the rubber, not the hole) and its inside diameter.

Measure the groove depth it will sit in.

Compute squeeze = (cord diameter − groove depth) / cord diameter × 100%. Aim for 15–25% for a static seal.

Fit a ring that is too thin for the groove — under about 5% squeeze — and pressurise the joint with the syringe.

Expect it to leak immediately.

Without initial squeeze there is no seal to pressurise, so the self-energising effect never starts. The ring must seal before pressure arrives in order to seal harder because of it.

Fit the correctly-sized ring — 15–25% squeeze — and pressurise gradually.

Expect it to hold, and to hold better as pressure rises rather than worse.

That is the whole invention in one observation. Most seals get worse with pressure; this one gets better, because the pressure is what pushes the rubber into the corner it must seal.

Now fit a ring that is much too fat for the groove — over about 40% squeeze.

It will seal. Then leave it compressed overnight and re-test.

Expect compression set: the rubber has taken a permanent flat and no longer springs back, so the seal degrades. Over-squeezing also raises friction badly on a moving seal and can tear the ring on assembly.

More is not better. There is a window, and both edges of it fail.

Set up a joint with a deliberately wide clearance gap beside the groove, fit a correct ring, and raise the pressure hard.

Expect the ring to be squeezed into the gap and to develop a shaved, nibbled edge — extrusion damage.

This is why an O-ring's pressure rating is meaningless on its own: it depends on the gap it is asked to bridge, the rubber's hardness, and whether a backup ring is fitted. Same ring, different housing, different limit.

The patent. US 2,180,795, "Packing", Niels A. Christensen, filed 2 October 1937, granted 21 November 1939. He began working on hydraulic seals in 1933 and spent four years testing rubber rings in slots. He was 72 when he filed.

What happened to him is worth stating plainly. Christensen licensed the patent to United Aircraft in 1941. After Pearl Harbor the United States government bought out key military patents and distributed them to manufacturers; Christensen received a lump sum of $75,000 and the O-ring effectively became public property. The most widely used seal in engineering made its inventor comparatively little. That is not a footnote — it is what the record shows, and a blueprint that told only the triumphant half would be misleading.

Why the groove is the component. An O-ring is a commodity; the gland it sits in is the engineering. Squeeze, gap, surface finish and groove fill are all specified, and a correct ring in a wrong groove fails. Design standards exist for the groove, not really for the ring.

The rubber is a separate decision. Nitrile for oil, EPDM for water and brake fluid, Viton for heat and fuel, silicone for extreme cold. Fitting the wrong compound looks identical and fails chemically — the ring swells, hardens or dissolves.

The most famous O-ring failure. The Challenger disaster in 1986 turned on O-rings in the solid rocket booster joints losing resilience at low temperature — they were too cold to spring back and seal the joint as it flexed at ignition. It is the clearest possible demonstration of step 4's lesson: a seal that cannot recover its shape is not a seal, and the material's temperature range is a specification, not a detail.