Pr1me also developed a F77 compiler fairly early on, although release times can be deceiving, as it took a long time before it was used in preference to their FTN IV compiler.
Their early versions of the F77 compiler suffered from both errors and poor performance/optimisation. (Most had this problem)
Early versions of the Microsoft PC Fortran compiler had similar characteristics, so was difficult to use.
This lead to alternative compilers, such as Silverfrost, Lahey, NAG etc. These had a number of features, including better reliability and diagnostics.
Lahey was early to provide a reliable F90 then F95 compiler.
It is very difficult to recall time lines for when these compilers appeared, then were reliable to use.
Extended and expanced memory also featured with the overlay environment, which is best forgotten.
It should also be remembered that CDC Fortran was a significant influence on Fortran IV usage, being the leader in numerical precision and moving to disk storage.
Hi Steve I started work at Imperial College Computer Centre in March 1978 and I wasn’t aware of anyone within the University of London using the Bell Labs Fortran compiler at that time.
Another Fortran IV project 
(Although this is a modern project, being written in Fortran IV.)