Open Data Stack Exchange is a question and answer site for developers and researchers interested in open data. It only takes a minute to sign up.
Sign up to join this community
In looking at the Census ACS data there are several columns that have line numbers ending in .5 or .7. Their table_ids are duplicates of ones immediately before or after, making some import more difficult than it should be.
They seem to not have corresponding data in the distributed CSV data, either.
Before seeing this, I actually verified this answer just earlier today with Paul Overberg, one of the leading Census journalists around: the "fractional line numbers" represent "headers": that is, they are labels that group together subsequent rows but which don't have values themselves.
His example was the "Median income in the past 12 months --" line in this: 
(from http://factfinder2.census.gov/bkmk/table/1.0/en/ACS/11_1YR/B07411)
