The 2006 election for U.S. House of Representatives District 13 in Sarasota County, Florida, attracted extensive controversy because an unusually high proportion of the ballots cast lacked a vote for that office, and the unusual number of undervotes probably changed the election outcome. Intensive technical studies based on examining software and hardware from the iVotronic touchscreen voting machines used to conduct the election failed to find mechanical flaws sufficient to explain the undervotes. Studies that examined the ballots used in Sarasota and in some other counties concluded the high undervote rate was caused by peculiar features of the ballot's format that confused many voters. I show that recorded events involving power failures and problems with the Personalized Electronic Ballots used with the machines correlate significantly with undervote rates in several Florida counties. The relationships between machine events and undervotes are sufficiently substantial and varied to make it unreasonable to discount the likelihood that mechanical failures contributed substantially to the high numbers of undervotes.