Doesn't matter... the point of the chart was that every time 20 was exceeded a p/e of 10 followed.
Sometimes it took years as was pointed out. The thing is that historically. valuations above the mean are corrected by a period of valuations under the mean.
Maybe... I remember a discussion on this very thread last year about the difference between a simple average of PE ratios and adding up the earnings and losses and calculating that way.