diff --git a/autofit/non_linear/mock/mock_samples.py b/autofit/non_linear/mock/mock_samples.py
index 14f89fe85..6ce8abf85 100644
--- a/autofit/non_linear/mock/mock_samples.py
+++ b/autofit/non_linear/mock/mock_samples.py
@@ -8,6 +8,22 @@ def samples_with_log_likelihood_list(log_likelihood_list):
     ]
 
 
+def prior_median_kwargs(model):
+    """
+    Placeholder sample values for a model, taken as the median of each prior.
+
+    A blanket value like ``1.0`` is invalid for any parameter whose prior
+    constrains it, so filling a mock sample with one builds an unphysical
+    instance. For example an elliptical profile's ``ell_comps`` must have a
+    magnitude below 1, and ``(1.0, 1.0)`` raises on construction. Prior medians
+    are always in range, so they are safe for every model.
+    """
+    if not model:
+        return {}
+
+    return {path: prior.value_for(0.5) for path, prior in model.path_priors_tuples}
+
+
 class MockSamples(SamplesPDF):
     def __init__(
         self,
@@ -40,7 +56,7 @@ def default_sample_list(self):
                 log_likelihood=log_likelihood,
                 log_prior=0.0,
                 weight=0.0,
-                kwargs={path: 1.0 for path in self.model.paths} if self.model else {},
+                kwargs=prior_median_kwargs(self.model),
             )
             for log_likelihood in range(3)
         ]
diff --git a/autofit/non_linear/mock/mock_samples_summary.py b/autofit/non_linear/mock/mock_samples_summary.py
index 509c9c493..a336b4d3a 100644
--- a/autofit/non_linear/mock/mock_samples_summary.py
+++ b/autofit/non_linear/mock/mock_samples_summary.py
@@ -1,4 +1,5 @@
 from autofit.mapper.prior_model.collection import Collection
+from autofit.non_linear.mock.mock_samples import prior_median_kwargs
 from autofit.non_linear.samples.sample import Sample
 from autofit.non_linear.samples.summary import SamplesSummary
 
@@ -23,7 +24,7 @@ def __init__(
 
         self._max_log_likelihood_instance = max_log_likelihood_instance
         self._prior_means = prior_means
-        self._kwargs = {path: 1.0 for path in self.model.paths} if self.model else {}
+        self._kwargs = prior_median_kwargs(self.model)
 
     @property
     def max_log_likelihood_sample(self):
diff --git a/autofit/non_linear/mock/mock_search.py b/autofit/non_linear/mock/mock_search.py
index dec1efd22..2a590b219 100644
--- a/autofit/non_linear/mock/mock_search.py
+++ b/autofit/non_linear/mock/mock_search.py
@@ -3,7 +3,7 @@
 from autofit import exc
 from autofit.graphical import FactorApproximation
 from autofit.graphical.utils import Status
-from autofit.non_linear.mock.mock_samples import MockSamples
+from autofit.non_linear.mock.mock_samples import MockSamples, prior_median_kwargs
 from autofit.non_linear.search.abstract_search import NonLinearSearch
 from autofit.non_linear.mock.mock_result import MockResult
 from autofit.non_linear.mock.mock_samples_summary import MockSamplesSummary
@@ -25,7 +25,7 @@ def samples_with_log_likelihood_list(log_likelihood_list, kwargs):
 
 
 def _make_samples(model):
-    return {path: prior.value_for(0.5) for path, prior in model.path_priors_tuples}
+    return prior_median_kwargs(model)
 
 
 class MockSearch(NonLinearSearch):
diff --git a/test_autofit/non_linear/samples/test_mock_placeholders.py b/test_autofit/non_linear/samples/test_mock_placeholders.py
new file mode 100644
index 000000000..521ee954b
--- /dev/null
+++ b/test_autofit/non_linear/samples/test_mock_placeholders.py
@@ -0,0 +1,63 @@
+import pytest
+
+import autofit as af
+
+
+class _RejectsUnitMagnitude:
+    """
+    Mirrors an elliptical profile's ``ell_comps`` guard: the pair is only
+    meaningful while its magnitude is below 1.
+    """
+
+    def __init__(self, ell_comps=(0.0, 0.0)):
+        if ell_comps[0] ** 2 + ell_comps[1] ** 2 >= 1.0:
+            raise ValueError(f"ell_comps magnitude is not below 1: {ell_comps}")
+
+        self.ell_comps = ell_comps
+
+
+def _guarded_model():
+    model = af.Model(_RejectsUnitMagnitude)
+    model.ell_comps.ell_comps_0 = af.UniformPrior(lower_limit=0.0, upper_limit=0.5)
+    model.ell_comps.ell_comps_1 = af.UniformPrior(lower_limit=0.0, upper_limit=0.5)
+    return model
+
+
+def test__mock_samples_summary__fills_with_prior_medians():
+    summary = af.m.MockSamplesSummary(model=_guarded_model())
+
+    assert list(summary.max_log_likelihood_sample.kwargs.values()) == [0.25, 0.25]
+    assert list(summary.median_pdf_sample.kwargs.values()) == [0.25, 0.25]
+
+
+def test__mock_samples__default_sample_list_fills_with_prior_medians():
+    samples = af.m.MockSamples(model=_guarded_model())
+
+    for sample in samples.sample_list:
+        assert list(sample.kwargs.values()) == [0.25, 0.25]
+
+
+def test__mock_result_without_samples_summary__builds_a_valid_instance():
+    """
+    A ``MockResult`` given a model but no ``samples_summary`` builds its own.
+    That placeholder used to fill every parameter with ``1.0``, which is an
+    invalid ``ell_comps`` — so reading ``.instance`` raised.
+    """
+    instance = af.m.MockResult(model=_guarded_model()).instance
+
+    assert instance.ell_comps == (0.25, 0.25)
+
+
+def test__all_ones_placeholder_would_be_rejected():
+    """
+    Guards the test above: confirms the model really does reject the old
+    blanket ``1.0`` fill, so it cannot silently stop testing anything.
+    """
+    model = _guarded_model()
+
+    assert model.instance_from_prior_medians().ell_comps == (0.25, 0.25)
+
+    with pytest.raises(ValueError, match="magnitude is not below 1"):
+        model.instance_for_arguments(
+            {prior: 1.0 for prior in model.priors_ordered_by_id}
+        )