Does this do what you want?

#!perl
use strict;

# First, group them by the first part of the numeric field
# and the other two columns
# then find how much of the rest of the numeric field is needed
# to distinguish the variations in the third column
my %grouping;
while(<DATA>) {
  my ($n, $n2, $c2, $c3) = /(\d+)\.(\S+)\s+(\w+)\s+(\w+)/;
  $grouping{$n}{$c2}{$c3}{$n2} = $_;
}
for my $n (sort {$a <=> $b} keys %grouping) {
  my $N = $grouping{$n};
  for my $c2 (sort keys %$N) {
    my $C2 = $N->{$c2};
    if (keys %$C2 > 1) {
#      print "$n$c2 is a complex group, containing:\n";
      for my $c3 (sort keys %$C2) {
        my $C3 = $C2->{$c3};
#        print "..(for c3=$c3):\n";
        for my $n (sort keys %$C3) {
          print "$C3->{$n}";
        }
      }
    }
    else {
      print "$n$c2 contains:\n";
      my $C3 = $C2->{(keys %$C2)[0]};
      for my $n (sort keys %$C3) {
        print "  $C3->{$n}";
      }
    }
  }
}
__DATA__
0024.118.3A FT TFF
0024.118.3B FT TFF
0024.118.3C FT TFF
0024.118.3D FT TFF
0024.118.4  FF TFF
0039.04     NN OTH
0039.205.2  FT OTH
0039.205.6  FT TFF
409.176.12A FT TFF
409.176.12B FT TFF
409.176.12C FT TFF
409.176.12D FT OTH
409.176.12E FT OTH
[download]