At 19:25 2004-10-14 +0200, Henric Nilsson wrote: >[...] I assumed that every call to plot erased the device before plotting. Please disregard! If it worked that way, the par(mfrow = c(2, 2)) would have been quite useless in the first place... Henric