Guide
When is a gauge accurate enough for a tolerance, and what does the 10:1 rule say about measurement uncertainty?
The 10:1 rule is the long-standing industry guidance that an instrument used to accept a dimension should resolve one tenth of that dimension’s tolerance or better. It is a quick screen for measurement uncertainty: if the gauge cannot see a tenth of the band, its readings near either limit are closer to a guess than to a measurement, and a tight tolerance on the drawing stops meaning much.
By Saola Manufacturing engineering team · Published · Updated
This guide is about the instrument, not the statistics. It explains which ratio people actually mean when they say “ten to one”, works out how fine a gauge each common tolerance needs, lists what adds to measurement uncertainty beyond the gauge itself, and shows how to write the gauge and decision-rule requirements into a drawing, RFQ and purchase order. Gauge R&R studies and capability indices are covered in the Cpk vs Ppk guide, and what tight tolerances cost to make is in the guide on the cost of tight tolerances.
On this page
Which ratio does the 10:1 rule actually set?
Three different ratios get called “the rule”: a 10:1 discrimination ratio between tolerance and gauge resolution, a gauge-maker’s allowance of a small fraction of the part tolerance for fixed-limit gauges, and a 4:1 ratio between tolerance and calibration uncertainty. They answer different questions, and a supplier can meet one while failing another.
The first ratio is about discrimination, or the smallest step an instrument can show. The AIAG Measurement Systems Analysis manual restates the old rule of ten as a rule of thumb and a minimum starting point for gauge selection: the instrument’s discrimination should divide the tolerance, or the process spread, into ten parts or more. A gauge that only reads in steps the size of a fifth of the tolerance can sort parts into “good” and “bad” but cannot show drift until a part is already out.
The second ratio is for go/no-go gauges, which have no readout at all. ASME Y14.43-2011 gives the gauge maker a small share of the product tolerance for making the gauge, plus an optional wear allowance, and under its absolute tolerancing policy places both inside the product limits, so a gauge that is within its own tolerance does not accept a part outside the product limits. The full treatment of plug and thread gauge classes is in the go/no-go thread gauge guide.
The third ratio comes from calibration. A test uncertainty ratio (TUR) of 4:1 compares the span of the tolerance with twice the 95 percent expanded uncertainty of the measurement. Per ANSI/NCSL Z540.3-2006, a US calibration standard that NCSL International has since withdrawn but that contracts still cite, the primary requirement is that the probability of falsely accepting an out-of-tolerance item does not exceed 2 percent; a TUR of at least 4:1 is the fallback where that probability cannot practicably be estimated. The standard covers the calibration of measuring equipment, and applying the same 4:1 figure to part inspection is a borrowed convention, not a requirement. Resolution is only one input to the uncertainty, which is why a 10:1 gauge can still fail 4:1.
How fine does the gauge need to be for a given tolerance?
Take the total tolerance band, divide by ten for the resolution the instrument must show, and divide by eight for the largest expanded uncertainty that still gives a 4:1 ratio. The table below does that arithmetic for common drawing tolerances and names the instrument class that typically clears both.
Use the total band, not the plus-or-minus value. A dimension toleranced at plus or minus one thousandth of an inch has a band of two thousandths, and the 10:1 discrimination target is a tenth of that band, not a tenth of the plus-or-minus value. The right-hand column is a starting point: an instrument class only qualifies once its calibration record and the conditions it is used in back up the number.
| Total tolerance band on the drawing | Resolution needed, 10:1 | Largest expanded uncertainty U (k=2), 4:1 | Instrument class that typically clears both |
|---|---|---|---|
| 0.020 in (0.51 mm) | 0.002 in (51 µm) | 0.0025 in (64 µm) | Digital caliper; height gauge on a surface plate |
| 0.010 in (0.25 mm) | 0.001 in (25 µm) | 0.00125 in (32 µm) | Caliper on clean, simple geometry; micrometer preferred |
| 0.004 in (0.10 mm) | 0.0004 in (10 µm) | 0.0005 in (13 µm) | Micrometer, bore gauge, gauge pins; not a caliper |
| 0.002 in (51 µm) | 0.0002 in (5 µm) | 0.00025 in (6 µm) | Tenths micrometer; dial bore gauge set to a master ring; go/no-go plug |
| 0.001 in (25 µm) | 0.0001 in (2.5 µm) | 0.000125 in (3 µm) | Bench comparator or air gauge set to a master, temperature controlled; a 1 µm digital micrometer is marginal at best |
| 0.0004 in (10 µm) | 0.00004 in (1 µm) | 0.00005 in (1.3 µm) | Air or electronic gauging in a temperature-controlled room |
Two patterns show up when you read down the table. The caliper drops out early, because a typical digital caliper resolves 0.0005 in but carries a stated accuracy of about 0.001 in either way, so on a bore with a total band of 0.002 in its own error uses the whole tolerance before anyone touches the part. And once the total band falls to about 0.001 in, the instrument is no longer the only hard part: at that level the master it is set to, the temperature of the part and the operator each take a share of the uncertainty budget comparable to the instrument’s own.
What goes into measurement uncertainty besides the gauge’s resolution?
Measurement uncertainty is the combined effect of every source of doubt in a reading: the instrument’s calibrated error, its repeatability, differences between operators, temperature, the form of the part and how the part is held. Resolution is often one of the smaller terms, not the largest.
The method for adding these up is set out in the international Guide to the Expression of Uncertainty in Measurement (JCGM 100:2008) and, in plainer form, in NIST Technical Note 1297. Each source is estimated as a standard uncertainty, the independent terms are combined as a root sum of squares (each weighted by how strongly it affects the result), and the result is multiplied by a coverage factor, usually k = 2, to give an expanded uncertainty with roughly 95 percent coverage. The terms that matter most on machined metal parts are these:
- Calibration and traceability. The instrument’s error against a reference, and the uncertainty of that reference, which only means something if the result relates to a stated reference, normally the SI as realized by a national metrology institute, through a documented unbroken chain of calibrations, as NIST describes metrological traceability.
- Repeatability and reproducibility. The scatter when one person measures the same feature repeatedly, and the shift between people. A gauge R&R study measures exactly this, and the NIST/SEMATECH chapter on measurement process characterization explains how.
- Temperature. Per ISO 1, drawing dimensions apply at 20 °C. Aluminum typically expands by about 23 µm per meter for each degree Celsius, so a 100 mm aluminum part that is 5 °C above 20 °C, measured with a gauge at 20 °C, reads about 12 µm long: about half of a total band of 0.001 in. If the gauge is just as warm, its steel grows about half as much as aluminum, which offsets part of the error but not all of it. The tight tolerance cost guide covers what holding the reference temperature adds to inspection cost.
- Part form. A two-point instrument on an out-of-round or tapered bore reports one diameter out of many. If the bore is lobed after turning or clamping, the reading depends on where the jaws land.
- Fixturing and datums. A height reading taken off a surface plate is only as good as the way the part sits on it. Position and profile need the datums simulated as the drawing defines them.
- Measuring force. Thin walls and soft alloys deflect under a micrometer’s ratchet or a caliper’s jaws, and the gauge reads the deflected size.
None of these show up in the resolution figure on the instrument’s spec sheet. That is why a 10:1 check on resolution is a necessary first screen, not proof that a measurement is fit for the tolerance.
What happens to a part measured right at the tolerance limit?
That depends on the decision rule, which is the agreed way of treating measurement uncertainty when accepting or rejecting a part. Without one, a reading just inside the limit can be accepted by the supplier and rejected by the customer, and both are right.
Under ISO 14253-1:2017, the default for GPS drawings is strict: the supplier proves conformance only inside the tolerance narrowed by a guard band at each end, and the customer proves nonconformance only outside the tolerance widened by a guard band. The 2017 edition sets that guard band from a default 95 percent conformance probability rather than from the full k = 2 expanded uncertainty U used in earlier editions; for a normal distribution it works out to 1.65 standard uncertainties, or about 0.83 U. Per ISO 14253-1:2017, a diameter with a total band of 50 µm (about 0.002 in), measured with an expanded uncertainty of 6 µm, can be accepted by the supplier only when the reading sits about 5 µm inside each limit, which leaves an acceptance zone of about 40 µm, and rejected by the customer only when a reading sits about 5 µm outside a limit. The strip between each tolerance limit and its acceptance limit is called a guard band.
ASME drawings carry no such default. ASME B89.7.3.1-2001 sets out what a stated decision rule has to cover but does not pick one for you, so on an ASME print the rule has to be stated in the contract or on the drawing. The ILAC-G8:09/2019 guidelines on decision rules describe the common options, including “simple acceptance”, where a reading inside the limit passes outright. Simple acceptance is reasonable when the uncertainty is small against the tolerance, which is the real point of the 4:1 ratio: with a good enough measurement, the guard band is too thin to argue about.
The practical consequence for a buyer is a trade. Guard-banding protects you, but it shrinks the supplier’s usable tolerance, and the cost of that shows up in the quote. If the band is already tight, it is often cheaper to ask for a better measurement than to tighten the acceptance zone.
Which gauge choices most often fail the 10:1 rule?
The usual failures are a caliper on a close-tolerance bore, a display resolution mistaken for accuracy, and a fine instrument used in conditions that throw away its precision. Each one produces a clean-looking report that cannot actually prove the dimension.
- Caliper on a bearing or press-fit bore. The caliper is the default tool at the machine and is fine for in-process trend checks. It is the wrong acceptance gauge for a fit diameter, for which the press-fit tolerance guide shows H7 hole bands of 15 to 25 µm per ISO 286-2 for diameters from 6 to 50 mm.
- Reading the display, not the spec sheet. An electronic indicator or micrometer can show one more digit than it can be trusted to. The figure that counts toward the 4:1 ratio is the stated accuracy or maximum permissible error plus everything in the previous section, not the last digit on the screen.
- A micrometer on a warm part. Parts measured straight off the machine carry heat from cutting. On aluminum, the temperature effect alone can exceed a tight tolerance, so close dimensions should be measured after the part has soaked to room temperature.
- Size gauges used for location. A pin gauge proves a hole’s size, not its position. Position needs a datum setup or a functional gauge; how bonus tolerance changes that check is in the MMC bonus tolerance guide.
- A gauge R&R study run on the wrong parts. A study run on a few setup pieces that barely differ makes the gauge look worse than it is when the result is compared with total variation, because the part-to-part spread it is judged against is missing; a study run by one person says nothing about differences between operators. The study should use production parts that span the expected process range and the people who will actually measure them, as the AIAG MSA manual describes.
- No record of which instrument was used. A first article report that lists a value but not the instrument cannot be checked against this table at all. The first article inspection guide lists the instrument column as a required field.
What should the drawing, RFQ and purchase order say about gauges?
Flag the few characteristics where measurement uncertainty matters, ask the supplier to name the instrument for each one before you place the order, and state the decision rule. General-tolerance dimensions do not need any of this.
Start on the drawing. Mark the critical characteristics with your company’s key-characteristic symbol and give them balloon numbers, so every later document can refer to them; the ballooned drawing guide covers the numbering. On a print that invokes the ISO GPS system, ISO 14253-1 applies by default, so add a note only if you want something other than its default rule. On an ASME print, a note such as the following removes the ambiguity:
MEASUREMENT DECISION RULE: SIMPLE ACCEPTANCE WHERE TUR ≥ 4:1; OTHERWISE GUARD-BANDED ACCEPTANCE PER ISO 14253-1.
In the RFQ, list the flagged balloons and ask three things for each: the instrument or gauge that will accept it, that instrument’s resolution and stated accuracy, and whether the reading is taken at the machine or after the part reaches room temperature. The answers tell you, before tooling, whether the supplier can prove the tolerance or only make it. Ask for a gauge R&R study only on the characteristics where the table above shows the gauge is close to its limit; the PPAP guide for non-automotive parts explains how to scale that request to a small order.
On the purchase order, turn the answers into a requirement. A clause along these lines is usually enough:
Characteristics marked as key on the drawing shall be accepted with instruments whose resolution is no coarser than 1/10 of the total tolerance and whose calibration is traceable to the SI through a national metrology institute. The first article report shall name the instrument used for each balloon.
Keep the clause to the flagged balloons. Applying it to every dimension on the print adds inspection time without adding information, which is the same trap the tight tolerance cost guide describes for over-tolerancing.
How are tight dimensions gauged at Saola?
The instruments in Saola’s published inspection plan and floor photos are calipers, go/no-go pin gauges, dedicated fixture gauges, a surface plate, a height gauge on a granite plate and go/no-go thread gauges. Every toleranced dimension is measured at first article, and a dimensional inspection record ships with each lot.
On CNC turning, ANPT and NPT threads are gauged to ANSI/ASME B1.20.1 and thread fit is checked with go/no-go gauges before a lot ships. For a die-cast base that is drilled and tapped after casting, the inspection plan published on the factory page pairs caliper checks with go/no-go pin gauges on every drilled hole, on every part, and checks flatness against a surface plate and cast bosses with a dedicated fixture gauge. When a drawing marks a dimension as a process-capability control point, that dimension is watched through the run, not only at first article, as the quality page describes.
The site does not list a coordinate measuring machine, air gauging, a temperature-controlled inspection room or gauge R&R studies. If the table above puts one of your characteristics in a row that needs them, say so in the RFQ and name the balloon, so the measurement method is settled in the quote rather than discovered at first article. You can also see the inspection bench on a live video walk of the factory before you commit.
Frequently asked
Is the 10:1 rule a requirement of ASME Y14.5 or ISO GPS?
No. Neither drawing standard sets a gauge ratio. The 10:1 figure is a guideline from measurement practice, restated in the AIAG MSA manual; it becomes a requirement only when a customer writes it into the purchase order or a quality agreement.
Is a 4:1 ratio the same as 10:1?
No. The 10:1 figure compares tolerance with resolution, the smallest step the gauge shows. The 4:1 figure compares tolerance with expanded measurement uncertainty, which includes calibration error, repeatability, temperature and the rest. A gauge can meet 10:1 on resolution and still miss 4:1 on uncertainty.
Do go/no-go gauges need to meet the 10:1 rule?
They have no resolution to compare, so the equivalent control is the gauge tolerance and wear allowance, which ASME Y14.43 keeps to a small share of the product tolerance and places inside the product limits. The GO member should be measured on a calibration schedule and retired when it reaches its wear limit, not when it starts passing bad parts.
Does a coordinate measuring machine automatically satisfy the rule?
Not automatically. A CMM’s uncertainty depends on the probe, the number and spread of points, the fixturing and the temperature, so the 4:1 check still applies to the specific feature. For a simple bore, a gauge set to a master ring can beat a CMM on repeatability.
Should I ask for gauge R&R on every dimension?
No. Ask for it on the key characteristics where the gauge is close to the tolerance, and accept the instrument name and calibration record for the rest. How R&R results feed into capability numbers is covered in the Cpk vs Ppk guide.
Sources
- Measurement Systems Analysis (MSA), 4th edition, AIAG, 2010
- Taylor and Kuyatt, NIST Technical Note 1297: Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, NIST, 1994
- JCGM 100:2008, Evaluation of measurement data — Guide to the expression of uncertainty in measurement (GUM), Joint Committee for Guides in Metrology, BIPM, 2008
- ISO 14253-1:2017, Geometrical product specifications (GPS) — Inspection by measurement of workpieces and measuring equipment — Part 1: Decision rules for verifying conformity or nonconformity with specifications, ISO, 2017
- ILAC-G8:09/2019, Guidelines on Decision Rules and Statements of Conformity, International Laboratory Accreditation Cooperation, 2019
- NIST/SEMATECH e-Handbook of Statistical Methods, Chapter 2: Measurement Process Characterization, NIST, 2012
- Metrological Traceability, NIST, updated 2024
- ASME Y14.43-2011, Dimensioning and Tolerancing Principles for Gages and Fixtures, ASME, 2011