I know the graph shown on the page and I think it is misleading. The amount of ethanol per sample is measured in ml per ml of sample, but the amount of methanol is measured in ml per ml of ethanol. Because the amount of ethanol is lower at the end of a run, the ratio of methanol per ethanol is higher. So in the graph it seems that the amount of methanol per sample is raised at the end of a run.
I suggest to take the graphs form the EU-study. Fig 1: behaviour of ethanol during the distilation Fig 2: behaviour of methanol (mg/l) during the distilation Fig 3 and fig 4 show the results of diverse calculations. And show why the tail of methanol/100 ml ethanol swishes upward at the end of the runs. Fig 5 and Fig 6 show the behaviour of total volatiles (incl methanol) and (excl methanol) in mg/l during the run.