Gauge blocks and roundness testers are not interchangeable calibration artifacts — a gauge block set certified to ±0.05 µm at 0.5–10 mm length (ISO 3650) cannot validate a roundness spindle, and a calibrated hemisphere cannot calibrate a height micrometer. Engineers often buy both because the measurands differ: length versus form.
National-lab fee schedules make the split explicit. NRC Canada's 2026 dimensional-metrology tariff lists short gauge block calibration (<100 mm) at $280 per block, length bars (>100 mm to 1000 mm) at $930, and roundness artifact work on a separate line with a $615 setup fee [S9]. Reference-grade gauge blocks (Pratt & Whitney grade AAA) hold ±0.05 µm at the measuring face [S4]. Labs running both length and form keep separate artifacts, separate uncertainty budgets, and separate certificate renewals — the comparison below is the shortcut most procurement specs skip.
Gauge Blocks Set Length; Roundness Testers Set Form
A gauge block is a rectangular length standard, wrung or stacked to produce a known dimension. ISO 3650 specifies a 30 mm × 9 mm cross-section for nominal lengths 0.5 mm to 10 mm and 35 mm × 9 mm for larger blocks, with grade-AAA reference blocks held to ±0.05 µm at the flat, parallel, length-measuring face [S4][S10]. A roundness tester is a spindle instrument that rotates a workpiece past a precision stylus or optical probe; it certifies a geometric deviation (roundness, cylindricity, concentricity) on a circular cross-section, not a length [S3][S5].
Inside a modern roundness instrument, the spindle is driven by a precision servo motor and the rotation index, datum centering, and stylus lift are sequenced by a PLC; neither component exists in a gauge block stack. The two artifact families share a lab, not a measurement axis, and they share almost no calibration equipment downstream.
Grade, Tolerance, and What the Certificate Actually Says
Pratt & Whitney's published grade ladder places reference (AAA) gauge blocks at ±0.05 µm tolerance, used as primary laboratory standards, and steps down through calibration, inspection, and workshop grades as tolerance widens [S4]. The grade is a length tolerance at the central measuring face, not a flatness or surface-finish rating, and it widens as block thickness increases inside the same grade [S4]. NIST's handbook (Monograph 180) frames the same point operationally: with grade-AAA blocks within 0.05 µm of nominal, the calibration report can be set aside and the nominal value used [S1].
Roundness testers do not carry a "grade" in the gauge-block sense; they carry a spindle accuracy spec — typically radial runout in µm or nm — and a stylus probe spec, both of which a calibrated artifact (hemisphere, ring, or plug) verifies rather than defines [S5][S9]. The artifact and the instrument are interdependent: a hemisphere with sub-0.05 µm form cannot certify a spindle whose own runout exceeds that deviation, so the artifact selection drives the detection threshold, not the other way around [S8].
Who Needs a Gauge Block vs a Roundness Tester
Buy a gauge block set if the production measurand is length, diameter (via stack plus sine bar), height, or step — typical when qualifying micrometers, calipers, height gauges, and CMMs [S7]. A gauge set is also the only practical length standard a field calibration tech can carry to a machine tool.
Buy a roundness tester if the production measurand is bearing race geometry, seal groove concentricity, shaft cylindricity, or any other form-tolerance callout on a drawing [S3][S5]. A bearing plant with a Class-6 form spec on the inner race needs form, not length. A pump-shaft shop with a 4 µm cylindricity callout needs the same.
Do not buy a roundness tester to qualify a caliper — a caliper's dominant error mode is slide straightness, not slide roundness, and a roundness instrument will not detect it. Do not buy a gauge block set to qualify a bearing race — race form is radial and out-of-plane, and a flat length stack has no sensitivity to it. Both are absolute limits: the measurand is wrong, and no tighter grade closes the gap.
Side-by-Side Comparison on Four Decision Criteria
Measurand, calibration cost, operating cost, and environmental sensitivity are the four criteria that separate gauge-block from roundness-artifact decisions, and a side-by-side comparison exposes why these two artifact classes cannot be merged into a single budget line. [S1]
1. Measurand. Gauge block = linear length (L). Roundness tester = form deviation (roundness, cylindricity, concentricity) on a circular cross-section. The two are orthogonal — neither substitutes for the other.
2. Calibration cost. NRC Canada's 2026 schedule lists short gauge block calibration (<100 mm) at $280 per block, length bars (>100 mm to 1000 mm) at $930, and roundness artifact calibration handled on a separate dimensional-metrology line item with a $615 setup fee [S9]. Plan the budgets separately.
3. Operating cost. Gauge blocks are passive: handle with gloves, wring on a precision surface, manage thermal soak. Roundness testers are active: spindle bearings, stylus tip wear, datum centering, filtered-air supply. Annual maintenance cost skews heavily toward the roundness instrument.
4. Environmental sensitivity. NIST documents that the difference between interferometric and mechanical gauge block length is visible at the sub-µm level when thermal soak is incomplete, and over 4,000 cross-comparisons between steel and chrome carbide masters make the drift a "considerable nuisance" [S2]. Roundness testers are also thermally sensitive but the dominant error is spindle runout, not block expansion [S8].
Real Failure Modes and Limitations
Gauge block thermal soak and roundness spindle bearing wear are the two failure modes a lab manager hits within the first year of operation, and both are documented in published metrology practice rather than vendor folklore. [S2]
First, gauge block thermal soak: a block handled with bare fingers for 10 s takes up to a few hours to return to nominal, and any deviation during that window biases the calibrated instrument downstream — Navy SBIR work (N221-009) flagged the same problem in defense metrology labs, where delays from thermal re-equilibration run into the hours and inflate calibration cost [S6]. Second, roundness spindle bearing wear: a spindle that has lost 0.5 µm of radial runout will pass a coarse artifact and fail a reference one, so the artifact selection — not the spindle spec — sets the detection floor [S8].
A gauge block cannot detect slide-straightness error in a caliper, and a roundness tester cannot detect flatness error in a surface plate. Both are absolute limits: the measurand is wrong, and a finer grade or tighter spindle spec cannot substitute.
Standards, Sourcing, and Cross-Calibration
ISO 3650 is the key document for gauge block cross-section and grade structure in the 0.5–10 mm and 35/47 mm length tiers, and it remains the procurement default for new sets [S10]. NIST Monograph 180 covers the interferometric-versus-mechanical comparison methodology and the ±0.05 µm reference tolerance at grade AAA [S1][S4]. Roundness artifact calibration traces to optical or stylus reference hemispheres, rings, and cylinders with certificates, not to a single ISO grade letter — a real difference that procurement paperwork tends to gloss over.
A full metrology cell that handles both length and form will typically also calibrate process instruments in the same facility — pressure sensors, flow meters, and industrial valves are common neighbours on the schedule, and labs often share the same environmental chamber. If you are sizing lab capacity, count the artifact, the technician hours, and the certificate renewal cost as separate line items — not as a single "calibration" budget.
Two trackable signals over the next 6–12 months. (1) NRC Canada and equivalent national labs typically post fee updates on an annual cycle, so the $280 short-block and $615 roundness setup fees visible on the 2026 schedule are the next benchmark when 2027 numbers publish. (2) ISO 3650 grade structure remains the procurement default for new gauge block sets, and any vendor substitution must publish an ISO 3650 cross-reference on the certificate — without it, the block is not traceable to a recognized standard.