Since VFP will do the necessary conversions to pass binary scalars to API routines, could you just multiply declare RtlMoveMemory with several aliases using VFP's DECLARE command, and have a different "aliased" version for each data type that you wanted to move somewhere? Like this:

DECLARE INTEGER RTLMoveMEmory IN win32API as VFPMoveDouble Double xSource, String @xDest, nBytes

DECLARE INTEGER RTLMoveMEmory IN win32API as VFPMoveFloat32 Single xSource, String @xDest, nBytes

So to move a Double:

xSource=3.141592
cTarget=Repl(" ",8)
VFPMoveDouble(xSource, @cTarget, 8)

And to move it as a single-precision float:

cTarget=Repl(" ",4)
VFPMoveFloat32(xSource, @cTarget, 4)

>Check this link:
>http://fox.wikis.com/wc.dll?Wiki~VFPFloatingPointDataType~VFP
>
>Link doubles together to a string buffer, or allocate memory and fill it with doubles. Depends on how is the external function declared.