diff --git a/implementations/racket/preserves/preserves/pexprs.rkt b/implementations/racket/preserves/preserves/pexprs.rkt
index 9686426..c902536 100644
--- a/implementations/racket/preserves/preserves/pexprs.rkt
+++ b/implementations/racket/preserves/preserves/pexprs.rkt
@@ -117,6 +117,7 @@
                               (match (peek-char in-port)
                                 [(or (? eof-object?) #\] #\> #\} #\)) *TRAILER-ANCHOR*]
                                 [_ (next)])))
+   #:interpreter-annotation (lambda (line) (record 'r (list 'interpreter line)))
    #:on-hash (lambda (in-port source next parse-error* default)
                (match-lambda
                  [#\{ (read-sequence 's in-port source next #\})]
diff --git a/implementations/racket/preserves/preserves/read-text-generic.rkt b/implementations/racket/preserves/preserves/read-text-generic.rkt
index d239a3f..ba16253 100644
--- a/implementations/racket/preserves/preserves/read-text-generic.rkt
+++ b/implementations/racket/preserves/preserves/read-text-generic.rkt
@@ -70,9 +70,10 @@
     [(_ re pat) (app (lambda (v) (regexp-try-match re v)) pat)]))
 
 (define ((make-preserve-text-reader #:reader-name reader-name
+                                    #:interpreter-annotation interpreter-annotation
+                                    #:read-annotated-value read-annotated-value0
                                     #:on-char on-char0
-                                    #:on-hash on-hash0
-                                    #:read-annotated-value read-annotated-value0)
+                                    #:on-hash on-hash0)
          in-port source read-syntax? decode-embedded0)
   (define read-annotations? read-syntax?)
   (define decode-embedded (or decode-embedded0
@@ -106,7 +107,7 @@
       [#\# (match (next-char*)
              [(or #\space #\tab) (annotate-next-with (read-comment-line))]
              [(or #\newline #\return) (annotate-next-with "")]
-             [#\! (annotate-next-with `#s(interpreter ,(read-comment-line)))]
+             [#\! (annotate-next-with (interpreter-annotation (read-comment-line)))]
              [#\f (unless (delimiter-follows?) (parse-error* "Delimiter must follow #f")) #f]
              [#\t (unless (delimiter-follows?) (parse-error* "Delimiter must follow #t")) #t]
              [#\" (read-literal-binary)]
diff --git a/implementations/racket/preserves/preserves/read-text.rkt b/implementations/racket/preserves/preserves/read-text.rkt
index 1099232..fb17b89 100644
--- a/implementations/racket/preserves/preserves/read-text.rkt
+++ b/implementations/racket/preserves/preserves/read-text.rkt
@@ -15,6 +15,7 @@
   (make-preserve-text-reader
    #:reader-name *reader-name*
    #:read-annotated-value (lambda (in-port source next parse-error*) next)
+   #:interpreter-annotation (lambda (line) `#s(interpreter ,line))
    #:on-hash (lambda (in-port source next parse-error* default)
                (match-lambda
                  [#\{ (sequence-fold in-port
diff --git a/implementations/racket/preserves/preserves/tests/samples.pr b/implementations/racket/preserves/preserves/tests/samples.pr
index 5bdb61b..ca05807 100644
--- a/implementations/racket/preserves/preserves/tests/samples.pr
+++ b/implementations/racket/preserves/preserves/tests/samples.pr
@@ -14,7 +14,7 @@
   "In each test, let stripped = strip(annotatedValue),"
   "                  encodeBinary(·) produce canonical ordering and no annotations,"
   "                  looseEncodeBinary(·) produce any ordering, but with annotations,"
-  "                  annotatedBinary(·) produce canonical ordering, but with annotations,"
+  "                  annotatedBinary(·) produce “canonical ordering”, but with annotations,"
   "                  decodeBinary(·) include annotations,"
   "                  encodeText(·) include annotations,"
   "                  decodeText(·) include annotations,"
@@ -39,6 +39,10 @@
   }>
   "Implementations may vary in their treatment of the difference between expectations"
   "21/22 and 31/32, depending on how they wish to treat end-of-stream conditions."
+
+  "The idea of canonical-ordering-with-annotations is to encode, say, sets with their elements"
+  "in sorted order of their canonical annotationless binary encoding, but then actually"
+  "*serialized* with the annotations present."
 ]>
 <TestCases {
   annotation1: <Test #x"85 B103616263 B00109" @"abc" 9>
diff --git a/implementations/racket/preserves/preserves/tests/test-pexprs.rkt b/implementations/racket/preserves/preserves/tests/test-pexprs.rkt
index 074d6a3..a90afeb 100644
--- a/implementations/racket/preserves/preserves/tests/test-pexprs.rkt
+++ b/implementations/racket/preserves/preserves/tests/test-pexprs.rkt
@@ -138,6 +138,16 @@ PrecType = "PREC" / "PREC_LEFT" / "PREC_RIGHT" / "PREC_DYNAMIC";
 ENDDOC
   )
 
+(define U #<<ENDDOC
+#!/example/of/a/shebang
+{
+    key: value
+    # example of a comment at the end of a dictionary
+}
+# example of a comment at the end of the input file
+ENDDOC
+  )
+
 (parameterize ((pretty-print-columns 80)) (write-pexprs (string->pexprs P #:read-syntax? #t)) (newline))
 (parameterize ((pretty-print-columns 10)) (write-pexprs (string->pexprs P #:read-syntax? #t)) (newline))
 
@@ -149,3 +159,5 @@ ENDDOC
 (parameterize ((pretty-print-columns 40)) (write-pexprs (string->pexprs S #:read-syntax? #t)) (newline))
 
 (parameterize ((pretty-print-columns 40)) (write-pexprs (string->pexprs T #:read-syntax? #t)) (newline))
+
+(parameterize ((pretty-print-columns 80)) (write-pexprs (string->pexprs U #:read-syntax? #t)) (newline))
diff --git a/preserves-expressions.md b/preserves-expressions.md
index c5d4d68..96fc623 100644
--- a/preserves-expressions.md
+++ b/preserves-expressions.md
@@ -3,7 +3,7 @@ title: "P-expressions"
 ---
 
 Tony Garnock-Jones <tonyg@leastfixedpoint.com>  
-February 2024. Version 0.3.1.
+May 2024. Version 0.3.2.
 
 [text syntax]: preserves-text.html
 
@@ -229,7 +229,7 @@ text-syntax encodings.
       # example of a comment at the end of a dictionary
   }
   # example of a comment at the end of the input file⌝
-= @<interpreter "/example/of/a/shebang">
+= @<r interpreter "/example/of/a/shebang">
   [ <b
         key <p |:|> value
         @"example of a comment at the end of a dictionary" <a>