diff --git a/Schema.Test/SchemaSemanticTypeTests.cs b/Schema.Test/SchemaSemanticTypeTests.cs
index 7cf785f..af00956 100644
--- a/Schema.Test/SchemaSemanticTypeTests.cs
+++ b/Schema.Test/SchemaSemanticTypeTests.cs
@@ -190,6 +190,51 @@ public void AnUnresolvedRefinementIsRejected()
 		Assert.Contains(i => i.Message.Contains("does not declare", StringComparison.Ordinal), schema.Validate());
 	}
 
+	/// <summary>
+	/// A chain of refinement ending at a name the schema does not declare is that one unresolved
+	/// name, and nothing else.
+	/// </summary>
+	/// <remarks>
+	/// Walking the chain stops both at a type already seen and at a link naming nothing, so the two
+	/// used to be indistinguishable to whoever asked: <c>A</c> was told it refined itself through a
+	/// cycle that does not exist, and the declaration to fix - <c>B</c>, which names the missing
+	/// type - was the one the message did not name. Depth one never reaches that branch, which is
+	/// why the case above did not catch it.
+	/// </remarks>
+	[TestMethod]
+	public void AnUnresolvedRefinementDeeperInAChainIsNotACycle()
+	{
+		Schema schema = new();
+		SchemaSemanticType weight = schema.AddSemanticType("Weight".As<SemanticTypeName>())!;
+		SchemaSemanticType force = schema.AddSemanticType("ForceMagnitude".As<SemanticTypeName>())!;
+		weight.SetUnderlyingType(new Semantic { SemanticTypeName = "ForceMagnitude".As<SemanticTypeName>() });
+		force.SetUnderlyingType(new Semantic { SemanticTypeName = "Missing".As<SemanticTypeName>() });
+
+		SchemaValidationIssue issue = schema.Validate().Single();
+
+		Assert.Contains("does not declare", issue.Message, StringComparison.Ordinal);
+		Assert.AreSame(force, issue.Element);
+		Assert.IsFalse(weight.RefinesItself());
+	}
+
+	/// <summary>
+	/// A cycle is still a cycle when the type that closes it is not the one asked.
+	/// </summary>
+	[TestMethod]
+	public void ARefinementChainThatRevisitsATypeStillReportsItself()
+	{
+		Schema schema = new();
+		SchemaSemanticType a = schema.AddSemanticType("A".As<SemanticTypeName>())!;
+		SchemaSemanticType b = schema.AddSemanticType("B".As<SemanticTypeName>())!;
+		SchemaSemanticType c = schema.AddSemanticType("C".As<SemanticTypeName>())!;
+		a.SetUnderlyingType(new Semantic { SemanticTypeName = "B".As<SemanticTypeName>() });
+		b.SetUnderlyingType(new Semantic { SemanticTypeName = "C".As<SemanticTypeName>() });
+		c.SetUnderlyingType(new Semantic { SemanticTypeName = "B".As<SemanticTypeName>() });
+
+		Assert.IsTrue(a.RefinesItself());
+		Assert.Contains(i => i.Message.Contains("refines itself", StringComparison.Ordinal), schema.Validate());
+	}
+
 	/// <summary>
 	/// A semantic type with no underlying type chosen is not generatable.
 	/// </summary>
diff --git a/Schema/Models/Schema.Validation.cs b/Schema/Models/Schema.Validation.cs
index 8935a2b..bc3f2f0 100644
--- a/Schema/Models/Schema.Validation.cs
+++ b/Schema/Models/Schema.Validation.cs
@@ -585,9 +585,12 @@ private void ValidateRefinement(Collection<SchemaValidationIssue> issues, Schema
 			return;
 		}
 
-		// Refines() stops at the first type it has already seen, so a cycle shows up as a chain
-		// that never reaches a non-semantic type rather than as a hang.
-		if (!semanticType.Refines().Any(refined => refined.UnderlyingType is not Semantic))
+		// Asked whether the chain reached anything real, a chain ending at a name the schema does
+		// not declare answers the same as a cycle, and the author is sent looking for one that is
+		// not there. RefinesItself() asks the question this message makes: whether the chain came
+		// back to a type it had already seen. The unresolved tail is reported at the declaration
+		// naming it, which is the declaration to fix.
+		if (semanticType.RefinesItself())
 		{
 			Report(issues, path, semanticType, $"Semantic type '{semanticType.Name}' refines itself, directly or through a cycle, so it is represented as nothing.");
 		}
diff --git a/Schema/Models/SchemaSemanticType.cs b/Schema/Models/SchemaSemanticType.cs
index 5bc3a76..344e26e 100644
--- a/Schema/Models/SchemaSemanticType.cs
+++ b/Schema/Models/SchemaSemanticType.cs
@@ -124,6 +124,27 @@ public IEnumerable<SchemaSemanticType> Refines()
 		}
 	}
 
+	/// <summary>
+	/// Gets whether the chain of refinement from this type comes back to a type already seen.
+	/// </summary>
+	/// <remarks>
+	/// <see cref="Refines()"/> stops for two reasons, and the chain looks the same from the outside
+	/// either way: a type it has already seen, which is a cycle, and a link naming something this
+	/// schema does not declare, which is not. Only the first is this type's to answer for — an
+	/// unresolved name is reported at the declaration that names it, which is where the fix is — so
+	/// telling the two apart is what stops a typo being reported as a cycle that does not exist.
+	/// <para>
+	/// The chain ended on a cycle exactly when the deepest type it reached still refines something
+	/// that resolves, since that is the link <see cref="Refines()"/> refused to follow a second time.
+	/// </para>
+	/// </remarks>
+	/// <returns><see langword="true"/> when the chain revisits a type.</returns>
+	public bool RefinesItself()
+	{
+		SchemaSemanticType deepest = Refines().LastOrDefault() ?? this;
+		return deepest.UnderlyingType is Semantic semantic && semantic.Declaration is not null;
+	}
+
 	/// <summary>
 	/// Gets the type values of this one are ultimately stored as, following any chain of
 	/// refinement down to the first type that is not itself semantic.