For sake of clarity we have color-coded this step-plot to distinguish its features. Notice that, the more accurately these observations are measured, the less likely you will observe two that are identical. In which case, a very large sample of very accurately measured observations would be almost smooth - which allows mathematical statisticians to simplify their calculations rather a lot.

The following code produces much the same sort of result using R.

Note, for comparison, we have produced two versions of the same plot.

wta=c(445,530,540,510,570,530,545,545,505, 535,450,500,520,460,430,520,520,430,535, 535,475,545,420,495,485,570,480,495,470,490) par(mfrow=c(1,2)) y=sort(wta) n=length(y); rr=1:n/n plot(c(y[1],y), c(0,rr), type='s', main='ECDF', xlab='Weight (kg)', ylab='Relative rank') points(y,rr, pch=20, col='red') plot(ecdf(wta), pch=20, main='ECDF', xlab='Weight', ylab='Relative rank')

Note:
• Because a stepplot is drawn in the order that values are presented to it, those values must be sorted beforehand.
• In order to get the scatterplot to draw the first vertical, we repeat the first y value (using c(y[1],y) and begin with a zero frequency using c(0,rr)

• The ecdf function does not need the data to be sorted beforehand, and passes the plot function virtually everything it needs.
• We have merely used the pch and col tags to modify the plot symbol to a small red dot - rather than the usual circle. If you want to omit the points altogether use plot(ecdf(y), pch = ' ')

• Somewhat irritatingly, plotting the output of R's ecdf function does not draw in any verticals, nor does it accept colour. Hence we usually prefer to produce our own ECDF plots.