For sets as mathematical objects, element order (as well as possible repetition) makes no difference. It "doesn't make sense" to talk about a set's "first" or "second" or "third", so the sets package's authors have not supplied a subsetting method that would allow you to retrieve indexed elements of "set"
-class objects.
To get a sense of why that's a good design decision that directly follows from the objects' faithful representation of sets' important mathematical structure, examine the following:
library(sets)
a <- set("item3", "item3", "item2", "item1")
b <- set("item1", "item2", "item3")
identical(a,b)
# [1] TRUE
If you insist on extracting set elements with a numeric index, you can always do something like this:
set(as.character(a)[1])
# {"item1"}
As for your second question, the outcome you're after can be got by using the enormously useful function do.call()
:
do.call(set, as.list(paste("item",1:3,sep="")))
# {"item1", "item2", "item3"}