Why no one thought of the concept of joins before the 70s?

So, let me see if I have my history right: - In 1976, the first system sold as an RDBMS was Multics Relational Data Store. I don't know if it had JOIN, but let's assume it did. - The UNIX command join was released in 1979. It imitates SQL's JOIN but it works on plain text files. So my question is. If join is relatively useful on plain text, why wasn't something like it developed before? Surely operating systems were dealing with plain text data and "tabular data" to some extent since the late 50s. And I'm sure at some point they would have found it useful to have a ready-made abstraction or tool to combine data from two different files in a meaningful way. So, why didn't the concept emerge before? Are my rose-tinted glasses underestimating the cognitive overhead of joins, with all their Venn diagramness, that is only justifiable when the performance of a RDB is needed? Were people content with merging data in an ad-hoc way? Or was join indeed a stroke of genius?