This was a famous mistake made by Lebesgue (see also Gerald Edgar's answer to this MO question).
Suslin showed that a plane Borel set of type exists whose projection is not a Borel set. See the references to the original article by Suslin and related works here.

