Here is the little set.seed experiment I mentioned in class.
The bottom line is that rnorm uses the "current" seed and then up-dates the seed.

Setting the seed once and doing rnorm(5) twice gives the same result as setting the seed and calling rnorm(10) once.

> set.seed(314159)
> rnorm(5)
[1]  0.04173265  0.78900901  1.00655630 -0.28952987 -1.14368338
> rnorm(5)
[1] -1.1242426  2.6205628 -0.4547841  0.5536584 -0.0305192
> set.seed(314159)
> rnorm(10)
 [1]  0.04173265  0.78900901  1.00655630 -0.28952987 -1.14368338 -1.12424259  2.62056279 -0.45478409  0.55365839 -0.03051920
>