Grab bananas of totally random weight — flat, no bell, pure chaos. Now average a handful at a time and plot those averages. As you average more per handful, a smooth bell curve emerges from the mess, every single time, no matter what the original mess looked like. This is the reason the bell curve is everywhere, and the reason polls of a few thousand can speak for millions. Slide the handful size and watch the bell sharpen.
The Central Limit Theorem: the sum (or mean) of N independent random variables with finite variance approaches a NORMAL distribution as N grows — regardless of the original distribution’s shape. If each variable has variance σ², the sample MEAN has variance σ²/N, so its spread shrinks like 1/√N — averaging 4× more data halves the scatter. This is why measurement error, and estimators built from sums, are so often bell-shaped, and it underwrites the standard error and confidence intervals of all of statistics. A fail-loud self-check throws unless the variance of the mean equals σ²/N. ◆ real probability, node-verified.
Requires independence and finite variance; heavy-tailed inputs (Cauchy) never converge, and strong dependence slows or breaks it. The 1/√N shrink of the mean’s spread is exact for any finite-variance source.