Although we don't see out-of-control points on the yearly average chart, the range chart for the yearly averages will contain out-of-control points. Also, the control limits on the Xbar chart are created using the mean range from the R chart. The UCL and LCL on the Xbar chart are inflated because of this. It's not surprising that the yearly average gas prices are in-control with respect to Rule 1 given an LCL of ~$1.70 and UCL of ~$4.20.

This data comes from a macro open system with a tremedous list of direct and indirect sources of variation. Stakeholders are trying to cause variation. What gives me a grin is that in the industrialized world we pay attention to gasoline price fluctations. We know what the price per unit volume is today. We have apps that tell us where the cheapest gas is. However, there are other bulk mateirals that cost way more per unit volume and we dont even try to know the variation. Why arent coffee prices advertised on signs? Why dont we know the daily price change in coffee (or milk or ....)? Is it the volume of consumption per unit of time and therfore the overall personal ecomomic impact? 

Just because we can crank numbers into a formula or software doesn't mean we should.  These observations are not independent by any stretch of the imagination.  Applying statistical limits to a process that by its very nature is heavily autocorrelated will lead to bad conclusions.  Wrong tools for this data.  We could change our sample frequency such that the autocorrelation disappears, but by that time there are very few data points to estimate limits properly.  Although I get that the article was to both teach SPC and promote PQ products, better data should have been chosen for that effort.