Ra is the arithmetic average of the absolute values of profile height deviations within the evaluation length. It tells you whether the surface is generally rough or smooth, but it cannot fully describe the shape of peaks and valleys. Two surfaces may yield similar Ra values: one with fine, uniform peaks and valleys; the other largely flat but containing a deep groove. Similar averages do not guarantee the same sealing performance. Once a deep groove crosses the sealing contact band, it acts like a water channel under a dam. Parker's O-ring handbook explicitly cautions that Ra alone is insufficient for comparing sealing surfaces; it must be combined with peak-to-valley depth, load-bearing area, and other information. Similar averages do not mean identical peak-and-valley morphology.

Roughness concerns short-wavelength micro-irregularities; waviness, flatness, and distortion are larger-scale issues. A stylus traversing a short distance may show a beautiful Ra, yet the entire flange may be slightly warped, or periodic waviness may remain from tool vibration. After assembly, the seal is compressed only at high points, while contact pressure at low points is insufficient. Once pressure and temperature rise, the medium penetrates through the weak zones. At this point, polishing the surface to a mirror finish often cannot cure the local pressure loss.

There is also an easily overlooked measurement trap: sampling length, cutoff wavelength, measurement direction, and position will change the result. Measuring along the machining marks versus perpendicular to them yields different information. A single short trace cannot represent the entire circumferential sealing band. Therefore, measurement conditions should be specified in the process and inspection requirements and kept consistent during re-testing. Beyond micro-roughness, large-scale form errors must also be examined.

Ra is a statistical parameter and may not reliably capture random defects. Handling damage, scratches dragged by abrasive particles, or tool withdrawal marks—if they cross the sealing band from the high-pressure side—can form a continuous leakage path. If the measurement location happens to avoid it, the report will still pass. Therefore, inspecting a sealing face cannot rely on a single "spot check" at a fixed position. After cleaning, examine the sealing band under low-angle light, focusing on through-scratches, indentations, porosity, and localized corrosion, then decide whether re-measurement or rework is needed. The most dangerous defects are often localized flaws that traverse the entire sealing band.

For rotary lip seals, directional machining marks on the shaft surface can produce a pumping effect. SKF therefore emphasizes reducing spiral marks on shaft surfaces. However, on certain static gasket flanges, manufacturers may instead specify controlled concentric or spiral serrations to help the gasket bite. This shows that "the smoother, the better" or "spiral marks are always wrong" are both inaccurate. The correct approach is to first distinguish between static and dynamic seals, sealing materials, media, and pressure, then execute the surface requirements specified in the drawings and by the seal supplier for that specific configuration.

First look for localized defects, then supplement with parameter and form checks.

A roughness tester does not provide a "no-leak" certificate—it provides partial information about a surface. Next time you encounter a leak despite an acceptable Ra, will you first check for deep grooves, waviness, or distortion of the entire surface?