A typical rel=canonical tag:
Often times, a page's content can be accessed via multiple URLs, especially if using URL parameters, or a group of pages are so similar it's only worth crawling or indexing one of them. The rel=canonical element is a way for the webmaster to identify which of these URLs is the representative from the group as the preferred page (also called the "canonical" page) and should be indexed.
For example, let's assume www.example.com/blue-widgets?sort-asc is a product listing page. www.example.com/blue-widgets?sort=desc is the same page with the sorting reversed. www.example.com/blue-widgets?sort=asc&source=ppc has the exact same content as the first page, but has an additional URL parameter to indicate where the traffic was referred from. If all three of these URLs were crawled normally, they would be flagged as being duplicate content.
Instead, if we put a rel=canonical tag on all three of these pages that said www.example.com/blue-widgets?sort-asc is the canonical version of the page, search engines would know to index this URL and ignore the other two, thus eliminating the duplicate content issue.