I used to have a similar problem, until Quigibo helped me.
First, you don't need to define a real var's size, it's always the same.

Second, when you reference a real var, you need to subtract 2 from the pointer.

• What is Zeros(9)→Z for? It shouldn't cause any problems, but unless you actually use it anywhere, you don't need it.
• Unlike most other variable types, float variables don't have a 2-byte header denoting their size. The GetCalc() routines unfortunately do not account for this, so they exhibit a number of problems when operating on float variables:
  
  • Using GetCalc() to either locate or create a float will expect a size header to be present and thus advance the pointer returned by 2 bytes. However, floats don't have this header and this advance shouldn't be made. This can be corrected for by subtracting 2 from the pointer returned from GetCalc() calls dealing with floats. Make sure to make this adjustment after checking whether or not the function was successful, though, so returned values of 0 indicating failure aren't changed to -2.
  • When creating a float, the GetCalc() routine will add an unwanted 2-byte size header. This effect can be negated by decreasing the size argument of the call by 2, so 9-2 for a real variable and 18-2 for a complex variable. I'm guessing you meant to create a real variable with GetCalc("varW",18) and just picked up the 18 from the memory management screen, but you really want 7. Unless you did mean to create a complex variable, in which case you want 16.