diff --git a/CHANGELOG.md b/CHANGELOG.md
index cf324dd1b18cb6a839df3a9010ddde7b080b0c7e..a3a80183c7b6a6980db3abfeb0abc9187fb9ea3e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,11 +8,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [Unreleased] ##
 
 ### Added ###
+
 - Identifiable class to represent the information used to identify Records.
 - Added some StructureElements: BooleanElement, FloatElement, IntegerElement, 
   ListElement, DictElement
 
 ### Changed ###
+
 - Some StructureElements changed (see "How to upgrade" in the docs):
   - Dict, DictElement and DictDictElement were merged into DictElement.
   - DictTextElement and TextElement were merged into TextElement. The "match"
@@ -27,6 +29,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed ###
 
+- [#39](https://gitlab.com/caosdb/caosdb-crawler/-/issues/39) Merge conflicts in
+  `split_into_inserts_and_updates` when cached entity references a record
+  without id
+
 ### Security ###
 
 ### Documentation ###
diff --git a/setup.cfg b/setup.cfg
index 5d76c9244bb21e2fd28a33e637be02dd7ef9fb6a..77c546f110c67e3c3a8f44cb617a7f1b187813e1 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -19,9 +19,9 @@ package_dir =
 packages = find:
 python_requires = >=3.8
 install_requires =
-    importlib-resources
-    caosdb > 0.9.0
-    caosadvancedtools >= 0.6.0
+	importlib-resources
+	caosdb > 0.10.0
+	caosadvancedtools >= 0.6.0
     yaml-header-tools >= 0.2.1
     pyyaml
     odfpy #make optional
diff --git a/src/caoscrawler/crawl.py b/src/caoscrawler/crawl.py
index 00b69aacb6aeff690ec860d3cc610138dc8026cd..e6b5f65cde7bd9159ab104f763597c3ded1c3903 100644
--- a/src/caoscrawler/crawl.py
+++ b/src/caoscrawler/crawl.py
@@ -29,36 +29,41 @@ the acuired data with CaosDB.
 """
 
 from __future__ import annotations
+
+import argparse
 import importlib
-from caosadvancedtools.cache import UpdateCache, Cache
-import uuid
-import sys
+import logging
 import os
+import sys
+import uuid
+import warnings
 import yaml
+
+from argparse import RawTextHelpFormatter
+from collections import defaultdict
+from copy import deepcopy
 from enum import Enum
-import logging
 from importlib_resources import files
-import argparse
-from argparse import RawTextHelpFormatter
+from jsonschema import validate
+from typing import Any, Optional, Type, Union
+
 import caosdb as db
+
+from caosadvancedtools.cache import UpdateCache, Cache
 from caosadvancedtools.crawler import Crawler as OldCrawler
-import warnings
+from caosdb.apiutils import (compare_entities, EntityMergeConflictError,
+                             merge_entities)
 from caosdb.common.datatype import is_reference
-from .stores import GeneralStore, RecordStore
-from .identified_cache import IdentifiedCache
-from .identifiable import Identifiable
-from .structure_elements import StructureElement, Directory
+
 from .converters import Converter, DirectoryConverter
+from .identifiable import Identifiable
 from .identifiable_adapters import (IdentifiableAdapter,
                                     LocalStorageIdentifiableAdapter,
                                     CaosDBIdentifiableAdapter)
-from collections import defaultdict
-from typing import Any, Optional, Type, Union
-from caosdb.apiutils import compare_entities, merge_entities
-from copy import deepcopy
-from jsonschema import validate
-
+from .identified_cache import IdentifiedCache
 from .macros import defmacro_constructor, macro_constructor
+from .stores import GeneralStore, RecordStore
+from .structure_elements import StructureElement, Directory
 
 logger = logging.getLogger(__name__)
 
@@ -759,7 +764,10 @@ class Crawler(object):
                 elif self.get_from_any_cache(identifiable) is not None:
                     # We merge the two in order to prevent loss of information
                     newrecord = self.get_from_any_cache(identifiable)
-                    merge_entities(newrecord, record)
+                    try:
+                        merge_entities(newrecord, record)
+                    except EntityMergeConflictError:
+                        continue
                     Crawler.bend_references_to_new_object(
                         old=record, new=newrecord, entities=flat + to_be_updated + to_be_inserted)
 
@@ -997,7 +1005,8 @@ class Crawler(object):
         if self.identifiableAdapter is None:
             raise RuntimeError("Should not happen.")
 
-        to_be_inserted, to_be_updated = self.split_into_inserts_and_updates(crawled_data)
+        to_be_inserted, to_be_updated = self.split_into_inserts_and_updates(
+            crawled_data)
 
         # TODO: refactoring of typo
         for el in to_be_updated:
diff --git a/unittests/test_issues.py b/unittests/test_issues.py
index ad66aa1413303a16b96ed877f3279a061a0a4bc5..ed30f7f619a6c054f09d35cc57a1746f01c664a6 100644
--- a/unittests/test_issues.py
+++ b/unittests/test_issues.py
@@ -22,15 +22,14 @@
 
 from pytest import mark
 
+import caosdb as db
+
 from caoscrawler.crawl import Crawler
+from caoscrawler.identifiable_adapters import CaosDBIdentifiableAdapter
 from caoscrawler.structure_elements import DictElement
 from test_tool import rfp
 
 
-@mark.xfail(
-    reason="Wait until value conversion in dicts is fixed, see "
-    "https://gitlab.com/caosdb/caosdb-crawler/-/issues/10."
-)
 def test_issue_10():
     """Test integer-to-float conversion in dictionaries"""
     crawler_definition = {
@@ -68,3 +67,46 @@ def test_issue_10():
     assert records[0].parents[0].name == "TestRec"
     assert records[0].get_property("float_prop") is not None
     assert float(records[0].get_property("float_prop").value) == 4.0
+
+
+def test_issue_39():
+    """Test for merge conflicts in
+    `crawl.Crawler.split_into_inserts_and_updates` (see
+    https://gitlab.com/caosdb/caosdb-crawler/-/issues/39).
+
+    """
+
+    crawler = Crawler(debug=True)
+
+    # For trying and failing to retrieve remotely identified records
+    def _fake_retrieve(*args, **kwargs):
+        return None
+
+    ident = CaosDBIdentifiableAdapter()
+    # identifiable property is just name for both Record Types
+    ident.register_identifiable("RT_A", db.RecordType().add_parent(
+        name="RT_A").add_property(name="name"))
+    ident.register_identifiable("RT_B", db.RecordType().add_parent(
+        name="RT_B").add_property(name="name"))
+    # overwrite retrieve
+    ident.retrieve_identified_record_for_identifiable = _fake_retrieve
+    crawler.identifiableadapter = ident
+
+    # a1 (has id) references b1 (has no id)
+    a1 = db.Record(name="A", id=101).add_parent(name="RT_A")
+    b1 = db.Record(name="B").add_parent(name="RT_B")
+    a1.add_property(name="RT_B", value=b1)
+
+    # a2 (no id) references b2 (has id)
+    a2 = db.Record(name="A").add_parent(name="RT_A")
+    b2 = db.Record(name="B", id=102).add_parent(name="RT_B")
+    a2.add_property(name="RT_B", value=b2)
+
+    flat_list = [b1, a1, a2, b2]
+
+    # the two records with ids exist remotely
+    crawler.add_to_remote_existing_cache(a1)
+    crawler.add_to_remote_existing_cache(b2)
+
+    # this would result in a merge conflict before
+    ins, ups = crawler.split_into_inserts_and_updates(flat_list)