Appendix B. The FORTRAN character set & collating sequence

The 49-character set

A standard FORTRAN 66 program is written entirely in these 49 characters (§3.1): the 26 letters, the 10 digits, the blank, and 12 special characters.

   letters   A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
   digits    0 1 2 3 4 5 6 7 8 9
   blank     (space)
   special   =  +  -  *  /  (  )  ,  .  $

Character	Name	Typical use
(blank)	blank	spacing; significant only inside Hollerith and in column rules
=	equals	assignment
+	plus	addition, unary plus
-	minus	subtraction, unary minus
*	asterisk	multiplication; ** exponentiation; * in column 1 = comment (forterp)
/	slash	division; record separator in FORMAT; common-block delimiters
( )	parentheses	grouping, subscripts, argument lists
,	comma	list separator
.	decimal point	in real constants; in .EQ., .AND., .TRUE., etc.
$	currency symbol	reserved by the standard; no standard meaning

Any other character (lower-case letters, :, ;, ', !, #, …) is not part of the standard FORTRAN character set. Such characters may appear only inside Hollerith data, where any processor-representable character — blank included — is allowed (§3.1.4.1, §4.2.6).

The collating sequence

FORTRAN 66 deliberately does not impose a collating sequence (§3.1, §3.6). The order in which the standard lists the characters carries no meaning, and the language provides no way to compare character (Hollerith) data with relational operators — relational expressions compare numbers only (Chapter 6). A program that needs to order characters must do so by its own arithmetic on the packed Hollerith values, and the result depends on the processor's internal representation.

forterp notes.

• forterp's host character representation is ASCII. On the default NATIVE target, Hollerith characters are packed into 64-bit words; on the faithful PDP10 target they are packed five 7-bit ASCII characters to a 36-bit word, exactly as DEC FORTRAN-10 did. Any program that compares or unpacks Hollerith data will therefore see different numeric values on the two targets — this is the value-model divergence discussed in Appendix C.
• Because forterp reads ASCII source, lower-case letters and characters like ! and ' are recognized where a dialect allows them (e.g. ! comments and '…' strings under FORTRAN10/ F77), even though they are outside the X3.9-1966 set. Under strict F66 they are rejected outside Hollerith, as the standard requires.
• A full ASCII table with the operator-precedence summary is in the FORTRAN 77 manual's Appendix C; the precedence rules there (arithmetic > relational > logical, with ** right-associative) apply unchanged to FORTRAN 66.