A gradient that is a function
The gradient of is 2 at x = 1, 4 at x = 2 and 6 at x = 3; at x = −2 it is −4, where the curve falls. At every x the gradient is 2x. So the gradient is not one number but a new function of x, made from the old one.
That new function is the derivative. It is used so often that it needs a short name, and two are in use: one from Lagrange and one from Leibniz. A third symbol, , is the instruction to find it.
The curve and the line y = 2x, its gradient function. Above x = 3 the curve has gradient 6, and the line has height 6. Above x = −2 the curve falls with gradient −4, and the line has height −4.
Lagrange: f'(x)
Name the function f, so . Its derivative is written f'(x), read "f prime of x". The small mark after the f is the prime, and it means the derivative of f. So f'(x) = 2x.
Put a number in and f' gives the gradient at that point: f'(3) = 2 × 3 = 6 and f'(−1) = −2. Compare f(3) = 9, the height of the curve at x = 3. The prime is the whole difference between a height and a gradient.
Each function keeps its own letter. If a second function is named g, its derivative is g'(x), and g'(4) is its gradient at x = 4.
Leibniz:
Leibniz worked with y and x. A change in x is written , read "delta x", and the change in y that goes with it is . The gradient of a chord is .
On , from (1, 1) to (2, 4), and , so . Shrink the steps and the chord turns toward the tangent at (1, 1). The limit of as tends to 0 is written , read "d y by d x". For , , so at (1, 1) it is 2.
is written like a fraction because it is the limit of one: a small rise over a small run. It is one symbol for that limit, not a division of a number called dy by a number called dx. and f'(x) name the same function, 2x, in two styles.
On , the chord from (1, 1) to (2, 4) has and , so . The less steep line is the tangent at (1, 1), where .
: an instruction
Written on its own in front of an expression, is an instruction: differentiate what follows, with respect to x. So is read "the derivative of with respect to x is 2x". When , is .
"With respect to x" names the variable that is changing. The letters can be others. If the area of a square of side r is , then , the rate the area grows per unit of side; and if , then , with as the instruction. The working is the same as for every time.
is not a number and nothing is multiplied by it. is not or ; it is the derivative of , which is 2x.
Differentiating twice
The derivative f'(x) = 2x is itself a function, so it has a derivative too. The graph of y = 2x is a straight line climbing 2 for every 1 across, so its gradient is 2 at every x. That is the second derivative of f, written f''(x), read "f double prime of x": f''(x) = 2.
In Leibniz's style the second derivative is , read "d two y by d x squared". It is applied to , and the two 2s count how many times has been applied: the 2 sits on the d at the top and on the dx at the bottom. For , .
Each prime and each 2 counts one more differentiation. f''(x) does not mean 2 × f'(x), which is 4x, and does not mean , which is .
The first derivative of is the line y = 2x, which climbs 2 for every 1 across. Its gradient is therefore 2 everywhere, and the level line y = 2 is the second derivative. The two cross at (1, 2).
The usual mistakes
Reading f(x) as the derivative. Without the prime nothing has been differentiated: f(3) = 9 is the height and f'(3) = 6 is the gradient. A capital F names a different function altogether.
Reading as . For at x = 3, , but . The gradient divides the small changes, not the values.
Turning upside down. is a small run over a small rise, which is 1 ÷ gradient.
Writing or for the second derivative. At x = 3, , but .
Reading f''(x) as f'(x) doubled or squared. Two primes mean differentiate twice: f''(x) = 2, while 2f'(x) = 4x.