**The most counterintuitive finding: *Iris-setosa* is the smallest flower on almost every measure, yet it has the widest sepals.** Sepal width does not grow with flower size, as you would expect. The species averages (from the `Iris_by_Species` gold table) show it: | Measure | setosa | versicolor | virginica | |---|---|---|---| | Average sepal width | 3.42 cm | 2.77 cm | 2.97 cm | | Average sepal length | 5.01 cm | 5.94 cm | 6.59 cm | | Average petal length | 1.46 cm | 4.26 cm | 5.55 cm | - **Sepal length and both petal measures** rise in order: setosa < versicolor < virginica. - **Sepal width** breaks the pattern. Setosa is widest (3.42 cm) and versicolor is narrowest (2.77 cm), with virginica in between (2.97 cm). Sepal width therefore orders the species differently from every other feature. **Why this matters for classification.** The min/max ranges show that the features differ in how well they separate the species: - **Setosa is cleanly separable on petals.** Its largest petal length is 1.9 cm, while the smallest in the other two species is 3.0 cm. Its largest petal width is 0.6 cm against a minimum of 1.0 cm for the others. A single threshold on petal size isolates it. - **Sepals are a poor separator.** Sepal width spans 2.3–4.4 cm in setosa, which overlaps the ranges of versicolor (2.0–3.4) and virginica (2.2–3.8). Sepal length overlaps heavily between versicolor (4.9–7.0) and virginica (4.9–7.9). - **Versicolor and virginica overlap on petals too.** Petal length runs 3.0–5.1 cm in versicolor and 4.5–6.9 cm in virginica. A simple rule will therefore not fully split those two, and a multivariate model is needed. The overlap claims come from per-species min/max ranges, not from counting misclassified rows. The evidence cannot say how many specimens actually fall in the overlap zones.