How does a derivative describe a rate at one moment?

A moving cart can be speeding up throughout a journey. Its total distance divided by total time gives an average, but that does not tell us how fast it is moving at one particular instant. A derivative answers the more local question by following what average rates approach as their time intervals become shorter.

Consider an invented cart motion with position f(t) = t² meters when the numerical time t is measured in seconds. More fully, the coefficient is 1 m/s², so multiplying by t² produces meters. At two seconds the cart is four meters from its starting point. At three seconds it is nine meters away. It travels five meters during that one-second interval, giving an average velocity of five meters per second. Because this cart keeps moving forward, its speed and the magnitude of its velocity agree.

A closer look uses the interval from two seconds to 2.1 seconds. The positions are four meters and 4.41 meters, so the average velocity is 0.41 divided by 0.1, or 4.1 meters per second. From two to 2.01 seconds, the same calculation gives 4.01 meters per second. Looking backward gives 3.9 over the interval from 1.9 to two seconds and 3.99 from 1.99 to two seconds. The two sides are approaching the same number: four.

The algebra explains the pattern. If the nearby time differs from two seconds by a nonzero increment h, the numerical difference quotient is [(2 + h)² − 2²]/h. Expanding the square gives (4h + h²)/h = 4 + h, in meters per second. As h approaches zero, the extra h approaches zero too. Every actual quotient still uses a nonzero interval. The limit asks where those quotients are heading; it never asks us to divide zero by zero.

On a position-versus-time graph, a line through two sampled points is a secant. Its rise divided by its horizontal run is the average velocity. As the second point approaches the first, the secant approaches the tangent at that point. For this curve at two seconds, the tangent's slope is four meters per second. The graph's upward curvature records the increasing velocity; the local tangent records its value at that moment.

The derivative also gives a useful nearby prediction. Starting at four meters at two seconds, continuing at four meters per second for 0.1 seconds would add 0.4 meters. That predicts 4.4 meters, close to the exact 4.41. Over only 0.01 seconds, the tangent predicts 4.04 meters, compared with the exact 4.0401. The discrepancy shrinks because the tangent describes the curve more closely over a smaller neighborhood.

Not every curve has a derivative everywhere. A sharp corner can give different limiting slopes from the left and right. Where a derivative does exist, it translates a changing history into a rate at one point: position into velocity, temperature into warming rate, or water level into rise per second. An instant has no duration, yet the pattern around it can still define a precise rate.