An alternative solution could be (keep in mind what @Tyler Rinker just said about replicate)

Data <- read.table(text='
312313.34
243444
12334.92
321312
353532', header=FALSE)

Data <- as.numeric(as.matrix((Data)))

set.seed(007)
Means <- replicate(1000, mean(sample(Data,50,replace=TRUE)))
Means consists of 1000 mean each one for every subsample of size 50. If you want the mean of the means do this:

mean(Means)
What you're trying to do sounds like a bootstrapping or something similar to resample techniques for bias reduction (I guess).