OK, following your messages on the chatterbox (I was not really on-line, I just had my browser still open on Perlmonks, but was not in front of my computer), I've just prepared a new version of the code that I supplied earlier.

I can't say if it does what you want, because I am still not even sure that I understood what you really want and because I don't have actual sample files on which to run tests, but at least, I tested that it should compile cleanly without any error or warning:

use strict;
use warnings;

my $file1=$ARGV[0];
my $file2=$ARGV[1];
my $file3=$ARGV[2];

open my $in1, "<", $file1 or die "unable to open $file1 $!";
open my $in2, "<", $file2 or die "unable to open $file2 $!";
open my $out, '>', $file3 or die "unable to open $file3 $!";

my $header1 = <$in1>;
my $header2 = <$in2>;
print $out "some header for output file \n";
# now you are all set with headers, you can focus on the data

while (my $line1 = <$in1>) {
     my $line2 = <$in2>;
     last unless defined $line2;
     chomp ($line1, $line2);
     my @values1 = split $line1;
     my @values2 = split $line2;
     my @values3 = map {$values1[$_] - $values2[$_]} 0..$#values1;
     print $out "@values3 \n";
}
[download]

This what it should do:

• read the first line of file1 and file 2 (the headers) and discard them;
• write a silly header in file3 (you can just change it in accordance with your needs);
• read into $line1 and $line2 the next line from file1 and file2 respectively;
• split $line1 and $line2 into the @values1 and @values2 arrays;
• substract respective values of @values2 from @values1;
• output the restults of these substractions to the output file;
• repeat the last four steps for the next line, until the files are exhausted.

So, if file1 has:

foo bar baz
43 41 67
[download]

And file2 has:

foo bar baz
41 2 60
[download]

The output would be:

Some new header to be determined
2 39 7
[download]

That is at least what I understood from your initial description of your requirement.

As I already said, this program would need additional error-handling code (if the files don't match), but I am not going to go into that so long as I don't even know if this is what you need.