The running means of these frequencies were calculated in the same way as you would calculate it for, say, daily rainfall data. In other words, the formula ignored the head capsule width, but assumed every head capsule width was represented by a number - even if that number was zero, as happened with the first point of this running mean (shown

red).

Lines joining these points were obtained by simple linear interpolation, and could enable you to estimate the probability of observing capsules of intermediate lengths. This would, of course, be inappropriate for genuinely discrete data.

Notice that, because these measurements are rounded to the nearest 0.025 mm, this variable behaves like a discrete variable. So point x is the mean of three frequencies: capsules x mm wide, capsules x-0.025 mm wide , and capsules x+0.025 mm wide.

These are part of the same data used in the introduction to Unit 3. But in that case, because we wished to show the entire frequency distribution, and larger head capsules were recorded to less accuracy, we were unable to use a simple 3-point running mean - and were forced to employ class intervals instead.