The differance of float and double dates back to a FORTRAN standard. In FORTRAN Variables aren't as loosly written as in PHP and you had to define variable types(OH NOES!). FLOAT or REAL*4 (For all you VAX people out there) defined the variable as a standard precision floating point, with 4 bytes of memory allocated to it. DOUBLE PRECISION or REAL*8 (Again for the VAX) was identical to FLOAT or REAL*4, but with an 8 byte allocation of memory instead of a 4 byte allocation.
In fact most modern variable types date back to FORTRAN, except a string was called a CHARACHTER*N and you had to specify the length, or CHARACHTER*(*) for a variable length string. Boolean was LOGICAL, and there weren't yet objects, and there was support for complex numbers(a+bi).
Of course, most people reading this are web programmers and could care less about the mathematical background of programming.
NOTE: Object support was added to FORTRAN in the FORTRAN90 spec, and expanded with the FORTRAN94 spec, but by then C was the powerful force on the block, and most people who still use FORTRAN use the FORTRAN77.