If that's all you need to do, your files are cleanly formatted, and you have a useful shell, you could do it in a one-liner:

$ perl -nlae "$t+=$F[2]; END{print $t/$.}" *
[download]

For a more flexible (and verbose) approach, the following could be a useful skeleton for your code:

use strict;
use warnings;
use File::Find;

sub process_file{
    return unless (-f);              # Only work on files, not directo
+ries

    open INPUT, '<', $_ or die "Couldn't open $_: $!";
    while (<INPUT>){                 # works through the file line-by-
+line
        chomp;                       # delete trailing new line
        my @columns = split /\s+/;   # split on whitespace

        print "TODO: Process the following columns: ",
               join(" -- " => @columns), "\n";
    }
    close INPUT or die $!;
}

# Find any file in a directory named "data", for example
# (and its subdirectories)
find(\&process_file, 'data');
[download]