tar ("tape archive") packs many files and directories into a single archive file, preserving their metadata — but on its own it does not compress; compression is a separate step.

tar glues a whole directory tree into one stream: file contents plus an index of names, sizes, permissions, ownership, and timestamps. That single-file result is easy to move around or hand to a backup, and because the metadata travels with it, an extracted tree is bit-for-bit the original (same permissions, same owners).

The thing to internalize: archiving and compression are two jobs.

• Archiving = many files → one file (what plain tar does).
• Compression = make that file smaller (what gzip/bzip2/xz do).

A bare .tar is therefore the same total size as its contents — it just has fewer files. You combine the two jobs (tar -czf) to get the familiar .tar.gz.

tar -cf archive.tar /path/to/files   # create: bundle, no compression
tar -tf archive.tar                  # list contents (test/table)
tar -xf archive.tar                  # extract

Heritage: the name comes from backing up to tape, which is sequential — hence an archive is a stream with an index, letting you pull out one file without reading the whole tape. That index is why individual extraction works.

Go deeper:

• GNU tar manual — archiving vs compression and the tape heritage.
• tar(1) man page — concise command reference.