diff --git a/scorm/README.txt b/scorm/README.txt
index 09d9aee..e4b6d55 100644
--- a/scorm/README.txt
+++ b/scorm/README.txt
@@ -10,7 +10,7 @@ In order to work with the SCORM API we first need a tracking storage.
   >>> tracks = TrackingStorage()
 
 We can now create a ScormAPI adapter. Note that this adapter is stateless
-as it is usually created anew upon each request.
+as usually a new instance is created upon each request.
 
   >>> from cybertools.scorm.base import ScormAPI
   >>> api = ScormAPI(tracks, 'a001', 0, 'user1')
@@ -31,7 +31,7 @@ Then we can set some values.
 
 Depending on the data elements the values entered are kept together in
 one track or stored in separate track objects. So there is a separate
-track for each interaction and one track for all the other elements.
+track for each interaction and one additional track for all the other elements.
 
   >>> for t in sorted(tracks.values(), key=lambda x: x.timeStamp):
   ...     print t.data
@@ -48,3 +48,17 @@ to care about the storage in different track objects.
   ('q007', '0')
   >>> api.getValue('cmi.interactions.1.result')
   ('incorrect', '0')
+
+We can also query special elements like _count and _children.
+
+  >>> api.getValue('cmi.objectives._count')
+  (0, '0')
+  >>> api.getValue('cmi.interactions._count')
+  (2, '0')
+
+  >>> api.getValue('cmi.interactions._children')
+  (('id', 'type', 'objectives', 'timestamp', 'correct_responses',
+    'weighting', 'learner_response', 'result', 'latency', 'description'), '0')
+  >>> api.getValue('cmi.objectives.5.score._children')
+  (('scaled', 'raw', 'min', 'max'), '0')
+
diff --git a/scorm/base.py b/scorm/base.py
index 5ba6484..4db5a44 100644
--- a/scorm/base.py
+++ b/scorm/base.py
@@ -31,6 +31,19 @@ from cybertools.tracking.btree import TrackingStorage
 
 OK = '0'
 
+_children = {
+    'cmi.comments_from_learner': ('comment', 'location', 'timestamp'),
+    'cmi.comments_from_lms': ('comment', 'location', 'timestamp'),
+    'cmi.interactions': ('id', 'type', 'objectives', 'timestamp',
+            'correct_responses', 'weighting', 'learner_response',
+            'result', 'latency', 'description'),
+    'cmi.learner_preference': ('audio_level', 'language',
+            'delivery_speed', 'audio_captioning'),
+    'cmi.objectives': ('id', 'score', 'success_status', 'completion_status',
+            'description'),
+    'score': ('scaled', 'raw', 'min', 'max'),
+}
+
 
 class ScormAPI(object):
     """ ScormAPI objects are temporary adapters created by
@@ -81,8 +94,18 @@ class ScormAPI(object):
 
     def getValue(self, element):
         tracks = self.storage.getUserTracks(self.taskId, self.runId, self.userId)
+        if element.endswith('._count'):
+            base = element[:-len('._count')]
+            if element.startswith('cmi.interactions.'):
+                return self._countSubtracks(tracks, base), OK
+            else:
+                data = self._getTrackData(tracks, '')
+                return self._countSubelements(data, base), OK
+        if element.endswith('_children'):
+            base = element[:-len('._children')]
+            return self._getChildren(base)
         prefix, key = self._splitKey(element)
-        data = self._getTrackData(tracks, prefix) or {}
+        data = self._getTrackData(tracks, prefix)
         if key in data:
             return data[key], OK
         else:
@@ -103,8 +126,26 @@ class ScormAPI(object):
         return '', element
 
     def _getTrackData(self, tracks, prefix):
-        track = None
         for tr in reversed(sorted(tracks, key=lambda x: x.timeStamp)):
             if tr and tr.data.get('key_prefix', None) == prefix:
                 return tr.data
         return {}
+
+    def _countSubelements(self, data, element):
+        result = set()
+        for key in data:
+            if key.startswith(element) and key not in result:
+                result.add(key)
+        return len(result)
+
+    def _countSubtracks(self, tracks, base):
+        return len([tr for tr in tracks if tr.data.get('key_prefix').startswith(base)])
+
+    def _getChildren(self, base):
+        if base.endswith('.score'):
+            base = 'score'
+        if base in _children:
+            return _children[base], OK
+        else:
+            return '', '401'
+