If you think of your R multiflora as an F1 hybrid, then selfing will give an F2 which may exhibit huge heterogeneity if the two parents combined to make the F1 were very different. So either the seedlings you are looking at “F2” came directly from out-crossing, or the plant you got them from “F1” was the product of a wide cross. Inbreeding reduces variability, outbreeding increases it.
Most of the time the F2 range between the two parents of the F1 for any particular trait. Of course what we are often looking for is the exceptions, where a mixing leads to greater expression or repression of one or another traits. The exception to this rule is when we try to combine, say yellow, from one rose (like Persian Yellow) with growth habit from another (a nice HT) and we want the final progeny to be similar to or in between the properties of the initial parents, e.g. a nice yellow HT.