My recollection, contrary to TFA, is that development of PL/I or PL1 (both names were used) started in 1964 or a bit earlier with the intention of coming up with a replacement for Fortran. I think some referred to it as Fortran V or Fortran VI. IBM introduced PL1 pretty quickly after the it rolled out 360 mainframes, maybe because it might help sell more megabuck hardware, but there were two different compilers that compiled two different versions of the language (a fork that remained a wretched stumbling block for at least 20 or 25 years) into the one for the big mainframes (model 50 or bigger) and one for the economical mid-line mainframes, and none for the model 20. In that period, Gerald Weinberg published some good stuff about PL1, including a strong warning against concurrent processing. There was also a PL1 on Multics that tried to be a little more suited to academia than the IBM versions. In the middle 1980's there was a PL1 subset that ran on PC's. It couldn't do anything that Turbo Pascal couldn't do, but it did it much slower.
The point of PL/I was to unify the userbases of both FORTRAN and COBOL, thus tending to "replace" both languages. There was some influence from PL/I to the CPL language, which in turn led to BCPL, B and C.
