diff --git a/README.md b/README.md
index f169199b..86c8d242 100644
--- a/README.md
+++ b/README.md
@@ -134,7 +134,7 @@ for d in bdd.pick_iter(u):
 n = bdd.count(u)
 # write to and load from JSON file
 filename = 'bdd.json'
-bdd.dump(filename, roots=[u])
+bdd.dump(filename, roots=dict(res=u))
 other_bdd = BDD()
 roots = other_bdd.load(filename)
 print(other_bdd.vars)
diff --git a/dd/_abc.py b/dd/_abc.py
index f60652a5..ce0dcde5 100644
--- a/dd/_abc.py
+++ b/dd/_abc.py
@@ -241,11 +241,12 @@ def dump(
 
         @type filename: `str`
         @type filetype: `str`, e.g., `"pdf"`
-        @type roots: container of nodes
+        @type roots:
+            - `list` of nodes, or
+            - for Pickle: `dict` that maps
+              names (as `str`) to nodes
         """
 
-    # revise this method to return roots as `list` or
-    # named nodes as `dict`
     def load(self, filename, levels=True):
         """Load nodes from Pickle file `filename`.
 
@@ -255,7 +256,11 @@ def load(self, filename, levels=True):
 
         @type filename: `str`
         @return: roots of the loaded BDDs
-        @rtype: `list` of nodes
+        @rtype: depends on the contents of the file,
+            either:
+            - `dict` that maps names (as `str`)
+              to nodes, or
+            - `list` of nodes
         """
 
     @property
diff --git a/dd/autoref.py b/dd/autoref.py
index bedf9aaf..886e63ce 100644
--- a/dd/autoref.py
+++ b/dd/autoref.py
@@ -240,7 +240,11 @@ def dump(self, filename, roots=None,
 
         @type filename: `str`
         @type filetype: `str`, e.g., `"pdf"`
-        @type roots: container of nodes
+        @type roots:
+            - `list` of nodes, or
+            - for JSON or Pickle:
+              `dict` that maps names (as `str`)
+              to nodes
         """
         # The method's docstring is a slight modification
         # of the docstring of the method `dd._abc.BDD.dump`.
diff --git a/dd/cudd.pyx b/dd/cudd.pyx
index a5c3679d..ec0e03e3 100644
--- a/dd/cudd.pyx
+++ b/dd/cudd.pyx
@@ -1822,7 +1822,11 @@ cdef class BDD:
 
         @type filename: `str`
         @type filetype: `str`, e.g., `"pdf"`
-        @type roots: container of nodes
+        @type roots:
+            - `list` of nodes, or
+            - for JSON:
+              `dict` that maps names (as `str`)
+              to nodes
         """
         if filetype is None:
             name = filename.lower()
@@ -1917,7 +1921,11 @@ cdef class BDD:
             where the BDD is loaded
         @type filename: `str`
         @return: roots of loaded BDDs
-        @rtype: `list` of `Function`
+        @rtype: depends on the contents of the file,
+            either:
+            - `dict` that maps names (as `str`)
+              to nodes (as `Function`), or
+            - `list` of `Function`
         """
         if filename.lower().endswith('.dddmp'):
             r = self._load_dddmp(filename)
diff --git a/examples/json_example.py b/examples/json_example.py
index a9acfbc5..305f575e 100644
--- a/examples/json_example.py
+++ b/examples/json_example.py
@@ -14,15 +14,16 @@ def dump_bdd_as_json(filename):
     bdd = _bdd.BDD()
     bdd.declare('x', 'y', 'z')
     u = bdd.add_expr(r'(x /\ y) \/ ~ z')
-    bdd.dump(filename, [u])
+    roots = dict(u=u)
+    bdd.dump(filename, roots)
     print(f'Dumped BDD: {u}')
 
 
 def load_bdd_from_json(filename):
     """Load a BDD from a JSON file."""
     bdd = _bdd.BDD()
-    u, = bdd.load(filename)
-    print(f'Loaded BDD: {u}')
+    roots = bdd.load(filename)
+    print(f'Loaded BDD: {roots}')
 
 
 if __name__ == '__main__':