Arithmetic mean
Center of the converted repeat set.
Unit Converters
Viscosity Precision Calculator
For viscosity repeatability studies, convert results obtained under the same method and condition, preserve every repeat, and report mean, sample standard deviation, relative standard deviation, and Type A standard uncertainty of the mean. Precision is kept distinct from accuracy and total measurement uncertainty.
Repeat measurements—
Mean viscosity—
Sample standard deviation—
Relative standard deviation—
Type A uncertainty of mean—
Observed minimum—
Calculation ledgerUnrounded values drive all decisions
| Repeat | Source result | Converted result | Deviation from mean |
|---|
How to use
Evaluate repeatability from measurements made under one controlled condition
Each entered number is one genuine repeat of the same viscosity measurand. The page preserves all repeats, converts them to one unit, and reports mean, sample standard deviation, RSD, and Type A uncertainty of the mean without calling repeatability accuracy.
- Collect at least two repeats using the same sample preparation and method.
- Hold temperature, pressure, spindle or shear condition, timing, and instrument configuration constant.
- Paste the ordered results and choose one common source unit.
- Review every row before interpreting the calculated spread.
- Keep calibration bias and non-repeatability uncertainty components outside this Type A calculation.
Repeatability fundamentals
Sample standard deviation
Observed spread estimated with n − 1 degrees of freedom.
RSD
Sample deviation divided by the absolute mean.
Type A of the mean
s/√n, describing repeat scatter in the estimated mean.
Accuracy distinction
A tight cluster can still share a systematic bias.
Result interpretation
Use s for repeat spread and uA for uncertainty in the mean
Sample standard deviation describes individual-repeat dispersion. Type A uncertainty becomes smaller with repeat count because it belongs to the mean. RSD is useful for relative comparison but becomes unstable when the mean approaches zero.
Statistical method
Convert first, then calculate statistics from every retained repeat
Unit conversion is applied row by row. The arithmetic mean uses all converted values, sample deviation uses n − 1, and Type A uncertainty divides s by √n.
Minimum count
One result has no sample standard deviation; additional repeats improve stability but do not cure bias.
Outlier handling
Do not delete a distant result merely to improve RSD; investigate and document an assignable cause.
Condition drift
Temperature or shear drift can create structured change that a single s value hides.
Condition stability
Separate drift from repeat scatter
A monotonic temperature or timing change creates structure that one standard deviation does not explain.
Repeat count
Interpret small-n statistics cautiously
Two repeats permit calculation but give one degree of freedom and an unstable spread estimate.
Outlier governance
Retain every result until cause analysis
A distant value may expose contamination, temperature loss, spindle slip, transcription error, or real heterogeneity.
Bias separation
Precision cannot reveal shared systematic error
Calibration, reference value, method, and operator bias require evidence outside the repeat set.
Reporting precision
Round only after the full calculation
Keep converted values and intermediate sums unrounded so display decimals do not alter s or RSD.
Visual explanation
Read every point before the mean and one-s band
The repeat strip preserves sequence and makes drift or isolated excursions visible. The mean line and ±s band summarize scatter; they are not specification or confidence limits.
Detailed calculation process
x̄ = Σxi/n; s = √[Σ(xi − x̄)²/(n − 1)]; RSD = 100s/|x̄|; uA(x̄) = s/√n
| Symbol | Meaning | Unit |
|---|---|---|
| xi | converted repeat | target viscosity unit |
| n | repeat count | count |
| x̄ | arithmetic mean | target unit |
| s | sample standard deviation | target unit |
| RSD | relative standard deviation | % |
| uA(x̄) | Type A standard uncertainty of mean | target unit |
Deviation check:
Default input and assumption register
Define the repeat set before calculating precision
The five defaults are separate repeat measurements under one declared method and condition. They are not a time trend, specification limits, or certified reference values.
| Input | Default | Statistical role | Control requirement |
|---|---|---|---|
| Repeat series | 0.997, 1.001, 0.999, 1.000, 1.003 mPa·s | sample observations | all valid repeats retained |
| Population | one repeatability condition | common source | same method and operator context |
| Spread estimate | sample standard deviation | n − 1 denominator | at least two observations |
| Type A result | s/√n | uncertainty of the mean | not total measurement uncertainty |
Secondary decision analysis
Separate scatter, relative precision, and uncertainty of the mean
The live diagnostics show what each statistic can support. Standard deviation describes individual-repeat scatter, RSD scales that scatter to the mean, and Type A uncertainty addresses the estimated mean; none of them measures systematic bias.
| Diagnostic | Current result | Interpretation | Follow-up |
|---|
Source evidence
Preserve the repeatability conditions and every accepted observation
Record sample preparation, instrument and geometry, method revision, temperature trace, pressure, spindle or shear rate, operator, sequence, timestamps, equilibration, cleaning, and raw instrument resolution. Any excluded observation needs a documented technical cause established before reviewing its effect on the statistics.
Limitations and consequences
Precision alone does not establish accuracy or conformity
This calculation excludes reference-value bias, calibration uncertainty, resolution, temperature uncertainty, between-day and between-operator reproducibility, sampling variability, and model effects. A small standard deviation can coexist with a large systematic error, and a few repeats can make the spread estimate unstable.
Viscosity-precision glossary
RepeatabilityPrecision under unchanged conditions.
Sample deviationEstimated spread of a sample.
RSDSpread relative to mean magnitude.
Type A evaluationUncertainty obtained statistically from observations.
AccuracyAgreement that also considers systematic effects.
ReproducibilityPrecision under changed conditions.
Additional viscosity-precision terminology
Degrees of freedomFor the sample standard deviation, the independent information count is n minus one after estimating the mean.
Standard errorEstimated standard deviation of the sample mean, calculated here as s divided by the square root of n.
Practical repeatability cases
Water reference check
Five repeats near 1 mPa·s give a small s, while calibration bias remains a separate uncertainty component.
Coating test
A spindle-speed change splits the observations into different conditions, so the laboratory does not combine them.
Decimal places are not proof of precision
Precision comes from controlled repeat data and transparent statistics, not from displaying more digits.
Viscosity precision FAQ
How many repeats are required?
At least two for s; more are normally needed for a stable estimate.
Why use n − 1?
The observed set estimates a broader repeatability population.
Is RSD accuracy?
No; it describes relative repeat spread.
Does Type A equal total uncertainty?
No. It is only the repeat-based contribution and excludes calibration, resolution, temperature, sampling, and method effects.
Should an outlier be deleted?
Only after documented cause analysis under the method.
Can temperatures differ?
Not within one controlled repeatability set.
What if the mean is zero?
RSD is undefined.
Can units differ by repeat?
No; normalize or separate the source results first.
Does adding decimals improve precision?
No. Additional display digits do not change measurement resolution, repeat scatter, or the information contained in the observations.
Advanced viscosity-precision questions
How many repeats are enough?
Two is the mathematical minimum, but rarely enough for a stable professional estimate. Choose the count from method requirements, expected variability, desired precision, and the decision risk. Choose the repeat count before viewing results and justify it from method requirements, expected scatter, decision risk, and desired precision.
Is Type A uncertainty the total uncertainty?
No. It includes only the statistical contribution estimated from these repeats. A complete budget may also include calibration, resolution, temperature, reference standards, sampling, and method effects. Build a separate uncertainty budget for calibration, resolution, temperature, sampling, and method effects before reporting total measurement uncertainty.
Can the sequence reveal drift?
The ledger and visual preserve order, so a trend may be visible, but this page does not fit or test drift. Investigate time dependence before treating all observations as exchangeable repeats. Plot and investigate sequence behavior when drift is suspected because the summary statistics alone assume exchangeable repeat observations.
Should an outlier be deleted automatically?
No. Investigate sample handling, instrument status, transcription, and method compliance first. Preserve the original record and the predeclared exclusion rule rather than deleting a point because it increases spread. Preserve every original result and the predeclared exclusion rationale so removing a point cannot silently improve the reported precision.
Why can RSD be misleading near zero?
RSD divides standard deviation by the absolute mean. When the mean is zero or very small, the ratio is undefined or unstable even when absolute repeatability is meaningful. Report absolute standard deviation whenever the mean approaches zero because RSD then becomes unstable or undefined despite measurable scatter.