About 27,100 people were required to vote by provisional ballot when they showed up at Philadelphia polling places last week - about double the number who were forced to use provisional ballots in the 2008 presidential election.

The figures reinforce complaints from the watchdog group Committee of Seventy about the relatively large numbers of would-be voters whose names were not listed in division poll books when they arrived at the polling stations.

In those situations, voters are not allowed access to the city's electronic voting machines.

Instead, they are permitted to vote via paper ballots. The ballots are sealed in separate envelopes with the voters' names, addresses, and signatures on the outside, to be held aside and counted later when voters' eligibility is confirmed through registration records.

Al Schmidt, who last week became cochairman of the three-member City Commissioners' panel, responsible for running city elections, provided the new provisional ballot figures Sunday as election workers prepared for their official count of the city's vote, which must be completed in a few weeks.

Schmidt said the commissioners' office would provide whatever resources were needed to get the provisional ballots counted in time to be part of the city's certified vote.

The figures made public after the polls closed Tuesday were unofficial, based on machine totals and not including provisional and absentee ballots.

Poll books are loosely bound lists of registered voters distributed in advance to each of the city's 1,687 voting divisions, theoretically identifying each registered voter in the division, along with his or her address and signature.

A supplemental list is also provided to cover voters whose registrations were processed too late to meet the deadline for printing the poll books.

There are a number of reasons voters might not find their names in poll books, starting with the simple explanation that they are not registered to vote - sometimes the result of poorly run voter-registration campaigns by third-party groups that collect thousands of applications but that do not check to make sure they are properly filled out, or that fail to turn in all the applications they receive. In such cases, voters might believe they are registered when they are not.

Schmidt cited two other possible factors for a difference between the 2008 and 2012 provisional totals:

Properly registered but confused voters showing up at the wrong polling place. More than a third of the city's divisions were moved to new locations between 2008 and 2012, Schmidt said, due mostly to efforts to comply with federal law on access for those with handicaps. The city sent postcards to registered voters in those divisions, but the notices weren't necessarily read.

A likely glitch in production of the 2012 poll books, which are printed by a private firm based on a computer extract of registered voters produced by the Pennsylvania Department of State.

Schmidt said he had contacted the state Election Bureau on Sunday to report a potential issue with voters who are properly registered in the state's voter-registration system, known as SURE, but who did not appear properly in poll books.