I'm not entirely sure what your question is, or where the confusion lies. Could you clarify?

As I said above, references are sort of like pointers in C, if you're familiar with that. A data structure lives somewhere in memory; another variable, a scalar (which might in turn be an element of an array or hash, of course) holds a reference to it. A redirect, if you will: "the data you're looking for is found in another location. Walk this way!"

Programming Perl has an entire chapter on references, so if you've got that book, I'd read that. (If you don't have that book, buy it, it's a must-have for any Perl programmer.) Also, see perlref for a less gentle introduction.

Anyhow, to sum it up:

• $row contains (colloquially: is) a reference to an array, i.e. the one that split returns.
• That array does not have a name of its own; i.e., it's not @array or any such thing.
• However, you can use the reference in $row -- "walk this way", as it were, following the directions it contains -- by dereferencing the reference. For an array reference, this is done by adding the array sigil, @: @$row.
• The same applies to $sheet2 and @$sheet2.
• Finally, @$sheet2 -- an array, remember! -- is used to store a list of references, each of which points to another array: one for each row.

So you have something like this for each line:

$row      @$row
+---+     +-----+-----+-----+--
| *-----> | ... | ... | ... | ...
+---+     +-----+-----+-----+--
[download]

and something like this for the entire file:

$sheet2   @$sheet2  (each of the @$row's)
          +---+
+---+     |   |     +-----+-----+-----+--
| *-----> | *-----> | ... | ... | ... | ...
+---+     |   |     +-----+-----+-----+--
          +---+
          |   |     +-----+-----+-----+--
          | *-----> | ... | ... | ... | ...
          |   |     +-----+-----+-----+--
          +---+
          | . |
            .
[download]

Does that make it clearer?