diff --git a/backend/fluksio/api/routes/runs.py b/backend/fluksio/api/routes/runs.py
index b7c52e0..123355e 100644
--- a/backend/fluksio/api/routes/runs.py
+++ b/backend/fluksio/api/routes/runs.py
@@ -146,6 +146,9 @@ class RunNodeRow(BaseModel):
     #: Which run it was restored from, when it was. That run is also where this
     #: node's series was recorded.
     cached_from: str = ""
+    #: That run has since been deleted, and this node's curve went with it. Its
+    #: outputs are still on this run's result.
+    curve_lost: bool = False
 
 
 class ArtifactRow(BaseModel):
@@ -781,6 +784,16 @@ def read_run(run_id: str, session: SessionDep) -> Any:
     ).all()
     detail = RunDetail.model_validate(run, from_attributes=True)
     detail.nodes = [RunNodeRow.model_validate(n, from_attributes=True) for n in nodes]
+    # Worked out here, not stored: a source run is deleted long after the
+    # nodes restored from it were written.
+    sources = {n.cached_from for n in nodes if n.cached_from}
+    kept = (
+        set(session.exec(select(col(Run.id)).where(col(Run.id).in_(sources))))
+        if sources
+        else set()
+    )
+    for row in detail.nodes:
+        row.curve_lost = bool(row.cached_from) and row.cached_from not in kept
     detail.artifacts = [
         ArtifactRow.model_validate(a, from_attributes=True) for a in artifacts
     ]
@@ -824,8 +837,8 @@ def _drop_runs(session: Session, chosen: Any) -> None:
     Two things this costs, both deliberate. ``RunNode.outputs`` *is* the stage
     cache, so a later run loses hits these would have served. And a node
     restored from one of them points here through ``cached_from`` — ``_series``
-    already reads a missing source as an empty curve, which is what ``NO_CURVE``
-    explains on the screen. The artifact bytes need no help: ``sweep_artifacts``
+    already reads a missing source as an empty curve, and ``read_run`` marks
+    such a node ``curve_lost``. The artifact bytes need no help: ``sweep_artifacts``
     keeps whatever a ``run_artifact`` row or a live message still names, so
     dropping the rows is enough and the hourly sweep reclaims the blobs.
     """
diff --git a/backend/tests/api/routes/test_runs.py b/backend/tests/api/routes/test_runs.py
index c543591..c982b9c 100644
--- a/backend/tests/api/routes/test_runs.py
+++ b/backend/tests/api/routes/test_runs.py
@@ -1108,6 +1108,11 @@ def test_a_curve_whose_run_is_gone_is_empty_rather_than_an_error(
     )
     assert answer.status_code == 200
     assert answer.json() == []
+    # And the run says why, rather than leaving an empty chart to explain itself.
+    (node,) = client.get(
+        f"{settings.API_V1_STR}/runs/reuse-1", headers=superuser_token_headers
+    ).json()["nodes"]
+    assert node["curve_lost"] is True
 
 
 def _metric(run_id: str, name: str, step: int, value: float, ts: float) -> RunMetric:
diff --git a/docs/concepts/runs.md b/docs/concepts/runs.md
index 7624362..75125d2 100644
--- a/docs/concepts/runs.md
+++ b/docs/concepts/runs.md
@@ -143,7 +143,8 @@ missing artifact bytes make an entry a miss.
 
 A cache hit does not replay emissions. Its metrics are read from the original
 run, so comparisons still show the recorded curve. Deleting that original run
-leaves cached results available but removes their curve.
+leaves cached results available but removes their curve. The run's page then
+names the deleted run each lost curve went with.
 
 ## Objects that cannot be serialized
 
diff --git a/frontend/src/client/schemas.gen.ts b/frontend/src/client/schemas.gen.ts
index 2851e61..1c3611c 100644
--- a/frontend/src/client/schemas.gen.ts
+++ b/frontend/src/client/schemas.gen.ts
@@ -3569,6 +3569,11 @@ export const RunNodeRowSchema = {
             type: 'string',
             title: 'Cached From',
             default: ''
+        },
+        curve_lost: {
+            type: 'boolean',
+            title: 'Curve Lost',
+            default: false
         }
     },
     type: 'object',
diff --git a/frontend/src/client/types.gen.ts b/frontend/src/client/types.gen.ts
index 270b8d5..b197d55 100644
--- a/frontend/src/client/types.gen.ts
+++ b/frontend/src/client/types.gen.ts
@@ -1206,6 +1206,7 @@ export type RunNodeRow = {
     logs: string;
     cache_key?: string;
     cached_from?: string;
+    curve_lost?: boolean;
 };
 
 export type RunPage = {
diff --git a/frontend/src/components/Runs/RunDetail.tsx b/frontend/src/components/Runs/RunDetail.tsx
index ac9f061..6a8b5c4 100644
--- a/frontend/src/components/Runs/RunDetail.tsx
+++ b/frontend/src/components/Runs/RunDetail.tsx
@@ -22,10 +22,10 @@ import { MetricPicker, RunMetricChart, useMetricNames } from "./MetricChart"
 import { OpenInDashboard } from "./OpenInDashboard"
 import {
   CARD,
+  curveLost,
   downloadArtifact,
   FLOW_GONE,
   isLive,
-  NO_CURVE,
   paramText,
   runQueryOptions,
   shortCommit,
@@ -53,10 +53,9 @@ export function RunDetail({ id }: { id: string }) {
   const nodes = run.nodes ?? []
   const artifacts = run.artifacts ?? []
   const reason = statusReason(run)
-  // A restored node's curve comes from the run it was restored from, so an
-  // empty chart here means that run is gone rather than that this one failed.
-  // Said once, where it applies.
-  const cached = nodes.some((node) => node.status === "cached")
+  // A restored node's curve comes from the run it was restored from, so it
+  // goes when that run is deleted. Said once, where it applies.
+  const lost = nodes.filter((node) => node.curve_lost)
 
   // What the flow declared when this run was submitted. A run from before
   // that was kept reads against the flow's current declarations.
@@ -206,8 +205,10 @@ export function RunDetail({ id }: { id: string }) {
         </section>
       )}
 
-      {names.length === 0 && cached && (
-        <p className={cn(CARD, "text-muted-foreground text-sm")}>{NO_CURVE}</p>
+      {lost.length > 0 && (
+        <p className={cn(CARD, "text-muted-foreground text-sm")}>
+          {curveLost(lost)}
+        </p>
       )}
 
       {Object.keys(run.result ?? {}).length > 0 && (
diff --git a/frontend/src/components/Runs/queries.ts b/frontend/src/components/Runs/queries.ts
index 4592682..eb186ec 100644
--- a/frontend/src/components/Runs/queries.ts
+++ b/frontend/src/components/Runs/queries.ts
@@ -285,6 +285,20 @@ export async function cancelThenDelete(runId: string) {
 export const NO_CURVE =
   "No curve was recorded. A node restored from the cache is read back from the run that produced it, so this draws nothing once that run has been deleted — its outputs are still on the result."
 
+/**
+ * Which restored nodes lost their curve, and with which deleted run.
+ *
+ * Named, since a chart that is merely short a few lines says nothing.
+ */
+export function curveLost(nodes: { node: string; cached_from?: string }[]) {
+  const runs = [...new Set(nodes.map((one) => shortId(one.cached_from ?? "")))]
+  const which =
+    runs.length === 1
+      ? `run ${runs[0]}, which was`
+      : `runs ${runs.join(", ")}, which were`
+  return `The curve of ${nodes.map((one) => one.node).join(", ")} was lost with ${which} deleted. A node restored from the cache reads its curve from the run that recorded it. Its outputs are still on the result.`
+}
+
 /** Why a run can name a flow that is not there any more. */
 export const FLOW_GONE =
   "The flow this ran has been deleted. The run is kept as the record of what ran, but there is nothing left to retry it against."