Change factor levels

Suppose I have a dataframe 'd' defined as

      L3 <- LETTERS[1:3]
      d0 <- data.frame(cbind(x = 1, y = 1:10), fac = sample(L3, 10, replace
= TRUE))
      (d <- d0[d0$fac %in% c('A', 'B'),])

   x y fac
2 1 2   B
3 1 3   A
4 1 4   A
5 1 5   A
6 1 6   B
8 1 8   A

Even though factor 'fac' in 'd' only has 2 levels, but it seems to bear the
birthmark of 3 levels from its parent 'd0':

str(d)

'data.frame': 6 obs. of  3 variables:
  $ x  : num  1 1 1 1 1 1
  $ y  : num  2 3 4 5 6 8
  $ fac: Factor w/ 3 levels "A","B","C": 2 1 1 1 2 1

How can I cut the umbilical cord so that factor 'fac' in 'd' would have an
accurate birth certificate with the correct number of levels? Apparently
the following does not work:

levels(d$fac) <- c('A', 'B')

Also any reason for this heritage?

Thanks,
Gang

	[[alternative HTML version deleted]]