Common Fortran Errors

P. A. Robinson

---

The vast majority of Fortran programming errors (not including trivial typos and syntax errors) fall under the following eight categories.

1. Wrong Placement of Text
   • If you run beyond the 72nd character on the line, the remaining characters will be ignored.

   • Placing continuation characters in other than the 5th column, or omitting them entirely causes continuation lines to be misinterpreted.

   • Omitting a comment character from the first column of a comment line causes the comment line to be interpreted as Fortran code.

   • Labels must be placed in the first 5 columns.

2. Wrong Variable or Function Type
   • This often occurs when variables are implicitly typed (either by you or using the default types, i-n = integer, a-h and o-z = single precision real). You need to ensure that the types of all functions and variables are declared wherever they differ from the implicit types.

   • This problem can occur when a typo causes the compiler to assume that you have introduced a new variable (e.g., you type the variable name ``variab1e'' instead of ``variable''; or, worse, ``variab le'' instead of ``variable'').

3. Decimal Point in Reals
   Be sure to use a decimal point at the correct location whenever
   • Inputting a real from the terminal or a data file. Otherwise it may be misread.

   • Using a real constant in the program. Otherwise it may be mistaken for an integer.

4. Double Precision Constants
   • Be sure to use double precision constants if you want your program to be truly double precision - you must write 1.d0, not 1., for example.

5. Integer Division.
   If you divide an integer by an integer the compiler assumes you want integer division with remainders discarded.
   • To ensure a real answer when dividing by an integer, divide by real("integer variable").

   • This problem often occurs when the decimal point is accidentally omitted when dividing by a real constant (cf., Category 3 above).

6. Ordering in Common Blocks
   • The order and type of variables in a common block must be the same wherever that block is used.

7. Mismatch of Numbers or Types of Variables in Subroutine/Function Calls
   • The number and type of variables in a call to a subroutine or function must match those occurring in the definition of the subroutine/function.

8. Writing Beyond an Array or Variable
   • If you assign values to array elements beyond the defined bounds of an array you may overwrite other variables (with bizarre results). This can also happen if you accidentally assign a complex, double precision, or real to an integer, or assign a double precision or complex to a real, for example. Don't forget that the array defined by ``array(n)'' consists of the elements 1 to n, not 0 to n-1.