A surface plate doesn't announce that it's gone out of tolerance. There's no warning light, no unusual sound, nothing that looks different to the naked eye. It just sits there looking exactly as flat as it did the day it was installed, while parts inspected against it drift quietly out of spec for reasons nobody in the shop can immediately explain. By the time someone thinks to recalibrate the plate itself rather than the gauge or the part, months of questionable inspection data may already be behind them.
This is one of the more overlooked maintenance gaps in a metrology lab or machine shop - plates get calibrated on installation and then, too often, left alone until something forces a recheck.
Wear isn't usually dramatic - it's localized and gradual
The most common cause of a surface plate going out of flatness isn't a single dramatic event. It's cumulative, localized wear from repeated use in the same zones. Parts get set down, slid, and measured in roughly the same areas of a plate over and over, and even though granite is extremely hard, it isn't infinitely wear-resistant. Over years of use, the center of a plate or the areas nearest common work zones can develop a shallow depression measured in microns - invisible to the eye, but enough to introduce measurable error in high-precision inspection work.
Loading patterns matter too. A heavy fixture or workpiece placed repeatedly in the same spot, especially if the plate isn't supported evenly underneath, can accelerate this kind of localized settling. It's a slower version of the same physics that causes any surface to compress unevenly under a concentrated, repeated load.
Thermal cycling causes movement even without wear
Separate from mechanical wear, a plate can also move due to thermal history, particularly if it wasn't fully stress-relieved before it was originally ground flat, or if it's been through significant temperature swings since installation. Stone that hasn't been given adequate time to age after quarrying can carry residual internal stress that releases gradually over years, showing up as a slow bow or twist that has nothing to do with how the plate was used and everything to do with how it was made.
This is part of why the aging and stress-relief process before final grinding matters so much for long-term stability, and why a manufacturer's claimed grinding tolerance on the day of shipment isn't the whole story - a plate that was flat when it left the factory but wasn't properly stress-relieved beforehand can still move meaningfully within its first year or two in service.
Damage that's easy to miss
Beyond gradual wear and thermal movement, physical damage is a more sudden but still often-overlooked cause of accuracy loss. Chips and dings at the edges are usually obvious, but the more insidious damage is subsurface - a plate that's been struck hard enough to create microfracturing beneath the surface without visibly chipping. This kind of damage can locally alter both flatness and the plate's response to future loading, and it typically isn't caught without instrumented inspection.
How recalibration actually gets done
Checking a surface plate's flatness isn't a single measurement - it involves mapping the plate across a grid pattern and comparing readings across that grid, usually using a precision level or an autocollimator-based method for larger plates, or a laser interferometer setup for the tightest tolerance requirements. Instruments commonly used for this kind of work include electronic precision levels capable of resolving fractions of an arcsecond, dial or lever-type comparators for local flatness checks, and laser interferometry systems for full-surface high-resolution mapping on the most demanding applications. The specific method and grid density typically follow the same standards referenced when the plate was originally graded - DIN 876, ASME B89.3.7, JIS B7513, and similar systems all specify how many measurement points a given plate size and grade requires for a valid calibration.
A reputable calibration should produce a full flatness map, not just a pass/fail statement, and the instruments used should carry their own calibration certificates traceable back to a national metrology institute. A plate that's simply stamped "calibrated" without an accompanying data map doesn't give you enough information to know whether it's marginally in tolerance or comfortably so - a distinction that matters a lot if you're planning to keep using that plate for another several years.
A practical recalibration interval
There's no single universal answer for how often a surface plate needs rechecking, since it depends heavily on usage intensity, load patterns, and environmental stability, but a plate in regular use in a working shop environment is commonly rechecked on an annual basis, with higher-use or higher-precision applications sometimes warranting more frequent verification. A plate that sits in a well-controlled, low-traffic metrology room with light, evenly distributed use can often go longer between checks - though "can" isn't the same as "should," and skipping recalibration to save cost is one of the more common ways a shop ends up with quietly unreliable inspection data without realizing it.
The underlying point is straightforward: a surface plate is a measuring instrument, not a piece of furniture. Treating it with the same calibration discipline applied to gauges and comparators - rather than assuming stone doesn't need attention because it looks unchanged - is what actually keeps the rest of a shop's measurement chain trustworthy.