During discussions of my previous hot tower analysis, concerns arose that the method wasn't going to find all hot towers. The previous method relied solely on a "bin percentage" method, where the counts in each bin were divided by the total counts of the histogram and the tower was counted hot if this ratio was greater than 0.2. 

The new method follows the same percentage scheme to find any super hot towers. But then we clone the tower hits spectrum, and mask out that tower from the spectrum. Then the average number of counts is calculated, and a tower is declared hot when it exceeds 10 times the average. Doing this in an infinte loop until there are no tower which exceed 10 times the average, any towers which could be hot towers are written out to text files.

Then with only these hot tower candidates in hand I run over the MuDst's and create histograms of the bad tower ADC, Pedestal, Gain and (ADC-Pedestal). These histograms for each tower are attached to this blog (for select runs from different fills), where I've grouped the histgrams for each tower on a single page for easy viewing.