You mention reading, so I assume this is a site with content meant to be read. The following studies examined the effect of line length on reading speed and on desirability. Dyson and Kipping conclude that for reading from a computer screen, reading speed increased with line length. Readers preferred the shortest and longest lines. Their study did not examine margins.
Youngman and Scharff tested similar conditions, but studied the effect of margins. They concluded that line length does not independently affect reading speed, but that line length and margin width interact significantly. Readers were fastest with 8-inch lines with 0 margin, but hated it. They strongly preferred 4-inch lines with 0.5- to 1.5-inch margins, and read nearly as fast as with 8-inch lines.
Youngman and Scharff (1998). Text width and margin width influences on Readability of GUIs. SWPA 1998: http://www.lieb.com/Readings/Width.pdf
Dyson, M.C. & Kipping, G.J. (1998). The effects of line length and method of movement on patterns of reading from screen. Visible Language, 32, 150-181. http://psychology.wichita.edu/surl/usabilitynews/72/LineLength.asp