Given: M20 concrete, four samples cast a week apart, sample results 26.4, 25.9, 27.1 and 26.0 N/mm² — each the mean of three cubes logged here with the same specimen ref and cast date. No established σ, so Table 8 gives σ = 4.0.
Group limit: 0.825 × 4.0 = 3.3 → rounded to the nearest 0.5 → 3.5, which exceeds the fck + 3 = 23 floor, so the group mean must be ≥ 23.5 N/mm².
Group mean = (26.4 + 25.9 + 27.1 + 26.0) / 4 = 26.35 N/mm² ≥ 23.5 → conforms, and every cube clears the individual floor of 20 − 3 = 17 N/mm².
Contrast: a week of poor control giving sample results 19, 18.5, 20, 19.5 → group mean 19.25 < 23.5 → the log flags the group FAIL and the rolling monitor drops with it — the moment to investigate batching and curing, well before the 28-day record closes.