What is left in each period when heating is a factor that can be reaosnably assumed to be attributable to the impact of heating and then anomalies in that factor looked for, but it gets screwy on that basis because weather has such a major influence.Looking at your data so far (for September each year) I see apparent anomalies 1998 over 1997, and 2009 (and since) over 2008 (and all years back to 1