Hello buchi2,

Welcome to the monastery.

It works almost as you have it, see sample of code bellow:

#!/usr/bin/perl
use strict;
use warnings;
use Data::Dumper;

sub dofn {
    my %hashfn;
    my ($ivon, $ibis, $imal) = @_;

    for (my $i = $ivon; $i < $ibis; $i++) {
    push(@{$hashfn{'var1'}}, $i*$imal);
    push(@{$hashfn{'var2'}}, $i*$imal*10.);
    push(@{$hashfn{'var3'}}, $i*$imal*20.);
    }

    return \%hashfn;
}

my $hierhash_1 = dofn(0, 5, 1);
# print "1 $hierhash_1->{'var2'}[3] \n";
push(my @hashfeld, $hierhash_1);

my $hierhash_2 = dofn(0, 5, 2);
# print "2 $hierhash_2->{'var2'}[3] \n";
push(@hashfeld, $hierhash_2);

# print Dumper \@hashfeld;

my $wert = $hashfeld[0]->{'var2'}[3];
print "$wert \n";

__END__

$ perl test.pl
30
[download]

Update: I modified a bit your code. I would suggest a few points to read and it is up to you how you will proceed. I would suggest not to use subroutines/Prototypes use simply subroutines/Pass by Reference. Why? A fellow monk haukex wrote a recent post about it Fun with Prototypes.

On Perl you do not need to pre-define your parameters you can use my straight on the parameter when you assign the value.

Last point to mention, I would return a reference (hashref, arrayref, complexref) instead of the hash. Why? Read more here Subroutines: Returning a hash vs. hash reference.

Hope this helps, BR.