To look for positive trends from PCs, I will throw out a couple of the low points in the data. There were some definite periods where the month was comprised of only a couple of rides (due to illness or travel). The periods thrown out are June and July 2004 (was going to Japan or Europe almost every other week) and September and October 2005 (was in Japan almost all of Sept without a bike, and October was like getting on the bike all over again).
To look for positive trends, I’ve taken the above data and averaged it across a yearly basis by month. So the May month is an average of the Mays from 2004 to 2008. The plots below show the average (thick black line) and 1-sigma, 2-sigma, and 3-sigma lines for the reported data. If PCs work, I expect my numbers to be above the 2-sigma line (thick yellow line). That would mark a small improvement in FTP, but after years of training where I feel I’ve reached a limit on FTP, even a 10 W gain would be significant (translating to 30-60 seconds quicker in a 40k).