PharmaCalc

X̄-R and X̄-S charts monitor a process sampled in rational subgroups — typically 3–5 items taken together at regular intervals: 5 tablets from a press every hour, 3 fill-weight checks per run. The X̄ chart tracks process location; the companion R (range) or S (standard deviation) chart tracks within-subgroup spread. Because control limits are built from within-subgroup variation only, the charts respond quickly to special causes between subgroups.

R or S? The range chart is the classical hand-calculation choice and works well for small subgroups (n ≤ 8 or so); the S chart uses every observation and is preferred for larger subgroups or software-computed charts. This calculator supports both and picks constants for any n from 2 to 25.

Typical applications: tablet compression monitoring, fill-weight sampling, dissolution with multiple units per timepoint, and CPV Stage 3 continued process verification.

Western Electric rules flagged by this calculator (applied to both charts): 1 — a point beyond ±3σ; 2 — 9 consecutive points on one side of the centerline; 3 — 6 consecutive points steadily increasing or decreasing; 4 — 2 of 3 consecutive points beyond ±2σ on the same side; 5 — 4 of 5 consecutive points beyond ±1σ on the same side. Each rule can be switched on or off to match your SOP. Naming note: Rule 2's 9-point run criterion follows Nelson (1984) — the default in most SPC software; the original Western Electric Handbook (1956) rule is 8 consecutive points, which is what PharmaCalc's I-MR calculator applies. The rule numbering also differs between the two calculators (here Rules 2–3 are the run/trend tests and 4–5 the zone tests; on I-MR, Rules 2–3 are the zone tests). Map your SOP to the rule definitions, not the rule numbers.

Scenario: Tablet compression monitoring, subgroup size n = 5, five subgroups of fill weights (mg). (Five subgroups keeps the arithmetic followable — production charts should have ≥ 20–25 subgroups before limits are fixed; until then treat limits as trial limits.)

1. Subgroup means and ranges.
   Data (mg)

   S1: 500.1, 499.8, 500.5, 500.0, 500.2 → X̄₁ = 500.12, R₁ = 0.7
   S2: 499.5, 499.9, 500.1, 500.3, 499.8 → X̄₂ = 499.92, R₂ = 0.8
   S3: 500.2, 500.8, 500.4, 500.0, 500.6 → X̄₃ = 500.40, R₃ = 0.8
   S4: 499.8, 500.2, 500.0, 500.5, 499.9 → X̄₄ = 500.08, R₄ = 0.7
   S5: 499.6, 500.1, 500.2, 499.9, 500.3 → X̄₅ = 500.02, R₅ = 0.7
2. Centerlines.
   Calculation

   X̿ = 2500.54 / 5 = 500.108 mg  ·  R̄ = 3.7 / 5 = 0.74 mg
3. X̄-chart limits (A₂ = 0.577 for n = 5).
   Calculation

   UCL = 500.108 + 0.577 × 0.74 = 500.53 mg
   LCL = 500.108 − 0.577 × 0.74 = 499.68 mg
4. R-chart limits (D₃ = 0, D₄ = 2.114).
   Calculation

   UCL_R = 2.114 × 0.74 = 1.56 mg · LCL_R = 0
5. Apply the rules. All five means fall inside [499.68, 500.53]; all ranges inside [0, 1.56]; no rule 2–5 pattern in five points. In control — and σ̂ = R̄/d₂ = 0.74/2.326 = 0.318 mg is the short-term σ available for capability work.

• Rational subgrouping: items within a subgroup must be produced under essentially identical conditions; mixing streams or time periods inside a subgroup hides the very signals the chart exists to find.
• Enough subgroups: limits computed from fewer than ~20–25 subgroups are trial limits and should be revised as data accumulates.
• Constant subgroup size: the constants assume equal n; changing n mid-chart requires recomputing limits.
• Independence and approximate normality of subgroup means: strong autocorrelation (e.g. continuous processes sampled rapidly) distorts limits — consider an EWMA chart instead.
• Control ≠ capability: an in-control process can still fail specification; follow with a capability study (Cpk/Ppk) once control is established.

• ASQ Certified Six Sigma Black Belt Handbook — control chart methods; Appendix 4 (control chart constants) and Appendix 6 (factors for estimating σ).
• Western Electric, Statistical Quality Control Handbook (1956) — the run rules.
• AIAG, Statistical Process Control reference manual — chart construction practice.
• NIST/SEMATECH e-Handbook of Statistical Methods, §6.3.2 — Shewhart charts.