Rainer pointed out, in response to my post, Row names in data frames: Beware of 1:nrow, that if I’d used rownames(x) <- as.character(1:3) rather than rownames(x) <- 1:3, I wouldn’t have had the problem I’d seen.

> x <- z <- data.frame(id=1:3)
> y <- data.frame(id=4:6)
> rownames(x) <- 1:3
> rownames(y) <- LETTERS[4:6]
> rownames(z) <- as.character(1:3)
> rbind(y,x)
  id
D  1
E  2
F  3
4  1
5  2
6  3
> rbind(y,z)
  id
D  1
E  2
F  3
1  1
2  2
3  3

If you type rownames(x) you see the same result as rownames(z), and is.character(rownames(x)) and is.character(rownames(z)) both return TRUE, but if you look at the "row.names" attribute directly, you see they are different.

> rownames(x)
[1] "1" "2" "3"
> rownames(z)
[1] "1" "2" "3"
> is.character(rownames(x))
[1] TRUE
> is.character(rownames(z))
[1] TRUE
> attr(x, "row.names")
[1] 1 2 3
> attr(z, "row.names")
[1] "1" "2" "3"

But why is 1:3 treated so differently from 2:4?

> w <- data.frame(id=1:3)
> rownames(w) <- 2:4
> attr(w, "row.names")
[1] 2 3 4
> rownames(rbind(y,w))
[1] "D" "E" "F" "2" "3" "4"