Distance is speed times time, and the same sentence is true of every rate: the change in a quantity is the rate of change times the elapsed time, provided the rate does not change. A tank filling at a steady 3 litres per minute gains 3·4 = 12 litres in 4 minutes. The proviso is the whole difficulty, because most rates worth asking about move.
Definition · Total change from a rate, in n piecesLet a rate f(t) be given on the interval from a to b. Cut the interval into n pieces of equal width h = (b − a)/n, with left ends a, a + h, a + 2h, …. The left sum is the total of rate times width, using the rate at each left end.
Lₙ = h·(f(a) + f(a + h) + f(a + 2h) + … + f(b − h)). It is exact when f is constant, and an estimate otherwise.
Take the rate 2t on the interval from 0 to 2: a tap that opens slowly, so that t minutes after it opens it is delivering 2t litres per minute. With n = 4 the width is h = 1/2, the left ends are 0, 1/2, 1, 3/2, and the rates there are 0, 1, 2, 3. Each piece contributes its rate times 1/2.
piece rate used rate · h
0 to 1/2 0 0
1/2 to 1 1 1/2
1 to 3/2 2 1
3/2 to 2 3 3/2
sum: 3
What the sum assumes
The sum is 3. The lab's exact-total tile prints 4, and the gap is not a rounding: the tap is still opening during every piece, and each piece used the smallest rate it contains. On a piece from 1 to 3/2 the rate climbs from 2 to 3, and the sum charged the whole piece at 2. The missing part of each piece is a small triangle of width 1/2 and height 1, of area 1/4, and four of them make the missing 1.
Where the total 4 comes from is the work of “The Antiderivative and the Fundamental Theorem”, later in this course. For now it is the number the sums are compared with, and the next lessons show them closing in on it. What matters here is the direction of the error: a rising rate and a left sum give a sum that is too low.
Example · Why the end rate times the elapsed time is not the totalA tempting shortcut for 2t on the interval from 0 to 2 is the rate at the end, 4, times the elapsed time, 2, which gives 8. That is twice the lab's total of 4.
The shortcut treats the rate as 4 for the whole two minutes, but it was 4 only at the very end. It was 0 at the start. Freezing the rate at its last value overstates the total by the same kind of error the left sum understates it, in the other direction and by more.
A steeper rate shows the same thing at a larger scale. The speed preset is t² + 1 on the interval from 0 to 3 in n = 3 pieces of width 1. The left ends are 0, 1, 2, the rates there 1, 2, 5, and Lₙ = 1·(1 + 2 + 5) = 8, against an exact total of 12. Three pieces are far too few for a rate that climbs from 1 to 10.
Raising the number of pieces helps, and the ramp shows it plainly: with n = 8 the left sum of 2t is 7/2, and with n = 16 it is 15/4. Both are still below 4, both are exact fractions, and neither is the total. How far the process can be pushed, and what it approaches, are the questions of the next two lessons.