diff --git a/docs/api.rst b/docs/api.rst
index 38e8e4c5..9fb2d129 100644
--- a/docs/api.rst
+++ b/docs/api.rst
@@ -67,7 +67,6 @@ Attributes
     Dataset.pr.comment
     Dataset.pr.institution
     Dataset.pr.history
-    Dataset.pr.source
 
 .. _dsmeth:
 
diff --git a/docs/data_format_details.rst b/docs/data_format_details.rst
index 433c8ac5..e20922d8 100644
--- a/docs/data_format_details.rst
+++ b/docs/data_format_details.rst
@@ -50,7 +50,6 @@ category-set (for example, ``animal (FAOSTAT)``).
 
 Additional rules:
 
-* The size of the ``source`` dimension must be exactly 1.
 * The valid values for the ``provenance`` are ``measured``, ``projected``, and
   ``derived``.
 
diff --git a/docs/data_format_examples.ipynb b/docs/data_format_examples.ipynb
index 0a3a52e2..42876f8b 100644
--- a/docs/data_format_examples.ipynb
+++ b/docs/data_format_examples.ipynb
@@ -201,7 +201,7 @@
     "    \"scenario (FAOSTAT)\": np.array([\"highpop\", \"lowpop\"]),\n",
     "    \"provenance\": np.array([\"projected\"]),\n",
     "    \"model\": np.array([\"FANCYFAO\"]),\n",
-    "    \"source\": np.array([\"RAND2020\"]),\n",
+    "    \"source\": np.array([\"RAND2020\", \"RAND2021\"]),\n",
     "}\n",
     "\n",
     "opulent = xr.Dataset(\n",
diff --git a/primap2/_data_format.py b/primap2/_data_format.py
index 09704fd2..8d859fba 100644
--- a/primap2/_data_format.py
+++ b/primap2/_data_format.py
@@ -385,10 +385,6 @@ def ensure_valid_dimensions(ds: xr.Dataset):
             f" sec_cats."
         )
 
-    if ds.dims["source"] != 1:
-        logger.error("Exactly one source required per data set.")
-        raise ValueError("Exactly one source required")
-
     for dim in required_indirect_dims.union(optional_indirect_dims):
         if dim in ds.attrs:
             split_dim_name(ds.attrs[dim])
diff --git a/primap2/_metadata.py b/primap2/_metadata.py
index 5664a189..964f0d74 100644
--- a/primap2/_metadata.py
+++ b/primap2/_metadata.py
@@ -72,12 +72,3 @@ def history(self) -> str:
     @history.setter
     def history(self, value: str):
         self._ds.attrs["history"] = value
-
-    @property
-    def source(self) -> str:
-        """a short identifier for the source of the data"""
-        return self._ds["source"].item()
-
-    @source.setter
-    def source(self, value: str):
-        self._ds["source"] = [value]
diff --git a/primap2/tests/data/test_sum_skip_allna_inhomogeneous_result.nc b/primap2/tests/data/test_sum_skip_allna_inhomogeneous_result.nc
index 4e20b427..8d079ad0 100644
Binary files a/primap2/tests/data/test_sum_skip_allna_inhomogeneous_result.nc and b/primap2/tests/data/test_sum_skip_allna_inhomogeneous_result.nc differ
diff --git a/primap2/tests/examples.py b/primap2/tests/examples.py
index 4a802467..8b5af8a4 100644
--- a/primap2/tests/examples.py
+++ b/primap2/tests/examples.py
@@ -48,7 +48,7 @@ def opulent_ds():
         "scenario (FAOSTAT)": np.array(["highpop", "lowpop"]),
         "provenance": np.array(["projected"]),
         "model": np.array(["FANCYFAO"]),
-        "source": np.array(["RAND2020"]),
+        "source": np.array(["RAND2020", "RAND2021"]),
     }
 
     # seed the rng with a constant to achieve predictable "randomness"
diff --git a/primap2/tests/test_data_format.py b/primap2/tests/test_data_format.py
index 61a96b5e..e32bd25c 100644
--- a/primap2/tests/test_data_format.py
+++ b/primap2/tests/test_data_format.py
@@ -20,10 +20,9 @@ def test_something_else_entirely(caplog):
     assert "object is not an xarray Dataset." in caplog.text
 
 
-def test_valid_ds_pass(minimal_ds, opulent_ds, caplog):
+def test_valid_ds_pass(any_ds, caplog):
     caplog.set_level(logging.INFO)
-    minimal_ds.pr.ensure_valid()
-    opulent_ds.pr.ensure_valid()
+    any_ds.pr.ensure_valid()
     assert not caplog.records
 
 
@@ -109,22 +108,6 @@ def test_wrong_dimension_key(minimal_ds, caplog):
     assert "'asdf' not in the format 'dim (category_set)'." in caplog.text
 
 
-def test_not_one_source(minimal_ds: xr.Dataset, caplog):
-    ds = minimal_ds.loc[{"source": "RAND2020"}]
-    ds["source"] = []
-    with pytest.raises(ValueError, match=r"Exactly one source required"):
-        ds.pr.ensure_valid()
-    assert "ERROR" in caplog.text
-    assert "Exactly one source required per data set." in caplog.text
-
-    caplog.clear()
-    ds["source"] = ["a", "b"]
-    with pytest.raises(ValueError, match=r"Exactly one source required"):
-        ds.pr.ensure_valid()
-    assert "ERROR" in caplog.text
-    assert "Exactly one source required per data set." in caplog.text
-
-
 def test_missing_sec_cat(minimal_ds, caplog):
     minimal_ds.attrs["sec_cats"] = ["missing"]
     with pytest.raises(ValueError, match="Secondary category 'missing' not in dims"):
diff --git a/primap2/tests/test_metadata.py b/primap2/tests/test_metadata.py
index 8c344c88..f05bd138 100644
--- a/primap2/tests/test_metadata.py
+++ b/primap2/tests/test_metadata.py
@@ -12,7 +12,6 @@ def test_metadata_properties(opulent_ds):
     assert ds.pr.history == (
         "2021-01-14 14:50 data invented\n" "2021-01-14 14:51 additional processing step"
     )
-    assert ds.pr.source == "RAND2020"
 
     ds.pr.references = "references"
     assert ds.pr.references == "references"
@@ -28,5 +27,3 @@ def test_metadata_properties(opulent_ds):
     assert ds.pr.institution == "institution"
     ds.pr.history = "history"
     assert ds.pr.history == "history"
-    ds.pr.source = "source"
-    assert ds.pr.source == "source"