FORTRAN came about first for scientific applications and later COBOL was developed for business.
C was the defacto language for Unix machines and was being used as a cross hardware platform assembly language as newer CPUs came on the market. At one time, Unix machines had a feature that would take C source code transmitted over the network, compile it and update the executable, saving it wherever you wanted which was subsequently disabled when a college student created one of the first worms that infected 6000+ machines (
http://en.wikipedia.org/wiki/Morris_worm).
C++ piggybacked on C (and literally too: where code was translated by the CFRONT program from C++ into C code and then compiled) introduced OO concepts to the UNIX world and became the preferred choice over Objective-C which was proprietary at the time and SmallTalk the first OO language.
SmallTalk while popular in OO research wasn't adopted for mainstream OO due to the limitation of coding in an IDE like environment where you could change anything in the SmallTalk core and where you had to export the whole workspace to someone else's machine for them to run your program. and they in turn had to import it and hope it didn't change something they had altered.
Given those choices FORTRAN, C and C++ survived to the present day.
So if you were writing a computer simulation of some physical problem you'd choose FORTRAN because of the large body of supporting libraries. Later as Unix got more popular you would choose C because it was available on more machines more so than FORTRAN. If you were a CompSci person you'd write it in C++ for greater modularity.
ADDENDUM: Matlab and Java are now taking over as the preferred languages for simulation. Matlab is primarily used by engineering students who then want to use it when they get a job (smart marketting on Matlabs part).
Java for many reasons: students learn it, its easier to learn than C++, runs cross-OS and cross-HW including Android HW has a lot of library support including the Open Source Physics library of methods and ODEs.