Canonical URL
Also known as: rel canonical, canonical link
A declaration in a page naming which URL is the real address for that content, so variants of it do not compete with each other.
Any page can be reached at more than one address: with and without a trailing slash, with tracking parameters attached, over http and https, with and without www. A canonical link tells a search engine which one to keep.
The usual form is self-referencing: the page declares itself as its own canonical.
A canonical is a hint, not an instruction, and search engines override it when the evidence disagrees. That is not a reason to skip it. It is a reason not to use it to paper over a structural duplication problem, which it will not fix.
The worst version of this failure is a canonical copied between templates that points every page at the homepage. It is silent, it looks correct in source, and it removes the entire site from the index except one URL.