The double derivative is to the derivative what the derivative is to the original function: it represents the RATE OF CHANGE of the derivative (like the derivative represents the RATE OF CHANGE of the original function). It is denoted by f''(x) and you calculate it by differentiating f'(x) (the derivative of f(x)).
The double derivative can be used (although of course it is never necessary in Methods) to verify the nature of stationary points, once you have used the derivative to find them.
For instance, if you find that for f(x), the derivative f'(x)=0 when x=3, x=5 and x=9, you can see that stationary points occur here. But in order to find out what type of stationary point they are, we need to either draw up a gradient table (this is what they teach you in Methods) OR calculate the double derivative at these points: f''(3), f''(5) and f''(9).
For instance, if f''(3) were to be positive, you could tell that the stationary point at x=3 is a local minimum, because while the gradient/derivative is zero at that point, the RATE OF CHANGE of the gradient (represented by the double derivative f''(3)) is positive. Intuitively, have a look at a local minimum turning point on any graph; see how the gradient starts off negative, then hits zero at the turning point, then becomes positive? That's because the rate of change of the gradient (the double derivative) is positive which causes the gradient to increase, and that's why we can use this test to determine that the stationary point is a local minimum. On the flipside, If f''(5) were negative, we could tell that there is a local maximum at x=5. If f''(9) were zero, there would be a stationary point of inflexion at x=9 (think of the point (0,0) on a x^3 graph).
Doing the double derivative test and drawing up a gradient table (to see what the gradient was before the stationary point and after) are two methods of determining what type of stationary point it was. You don't have to use the former in Methods.