diff --git a/KOHEN.md b/KOHEN.md
index ca49e80d..7dc4679c 100644
--- a/KOHEN.md
+++ b/KOHEN.md
@@ -55,7 +55,7 @@ You inherit infrastructure that has been provisioned and verified:
 | cert-manager + nginx-ingress | Running. ClusterIssuer `letsencrypt-prod` working. |
 | Gitea (persistent storage) | Running. OAuth app + PAT exist for the operator's account. |
 | Gitea MCP server | Running at `https://mcp.shorashim.ch`. Connected to claude.ai. |
-| L1 Torah filesystem | In the `shorashim` repo at `l1/torah/`. SHA-256 `f743b34b…3f1494` (full hash in `l1/integrity.sha256`; verified at Day 1). |
+| L1 Torah filesystem | In the `shorashim` repo at `l1/torah/`. SHA-256 `9e1abe4e…3c3086` (full hash in `l1/integrity.sha256`; verified at Day 1). |
 | Image pull secret `shorashim-registry` | **Verify before assuming.** A prior partial setup may have created it. `kubectl get secrets -n shorashim` is the source of truth. |
 | `shorashim-secrets` (DATABASE_URL) | **Verify before assuming.** Same as above. If present, also verify the role it points at (the architecture requires `shorashim_app`, not `avnadmin`). |
 | `shorashim-config` ConfigMap | **Verify before assuming.** Runtime config (Shabbat lat/lng, paths, timeouts) — Construction §4.4. |
@@ -105,7 +105,7 @@ cd ~/shorashim   # or wherever you cloned
 # 1. The scroll itself hashes to the constant baked into the binaries.
 #    (Until 2026-10-04 this step compared integrity.sha256 with the
 #    constant — two copies of one string. This reads every verse.)
-expected="f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+expected="9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 python3 scripts/verify_l1.py l1 "$expected"
 # Expect: computed = recorded = expected, "OK: the scroll matches"
 
diff --git a/KOHEN_PROMPT.md b/KOHEN_PROMPT.md
index dd1d40a0..a71cdd4d 100644
--- a/KOHEN_PROMPT.md
+++ b/KOHEN_PROMPT.md
@@ -77,7 +77,7 @@ test -d ~/shorashim/specs && test -d ~/shorashim/l1/torah && echo "repo OK"
 
 # 0d. L1 integrity matches the firmware constant
 cd ~/shorashim
-expected="f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+expected="9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 python3 scripts/verify_l1.py l1 "$expected" || echo "L1 FAIL — STOP"
 
 # 0e. Specification identity (Doorpost)
diff --git a/azazel_processor/src/azazel_processor/goral.py b/azazel_processor/src/azazel_processor/goral.py
index 5952a773..46797c75 100644
--- a/azazel_processor/src/azazel_processor/goral.py
+++ b/azazel_processor/src/azazel_processor/goral.py
@@ -77,7 +77,7 @@ from enum import Enum
 # the hash changes, this constant updates too. The shared value is
 # how the goral's seed inherits L1's tamper-resistance.
 _L1_INTEGRITY_HASH = (
-    "f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+    "9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 )
 
 
diff --git a/l1/README.md b/l1/README.md
index 220b759c..d2b3346d 100644
--- a/l1/README.md
+++ b/l1/README.md
@@ -90,7 +90,40 @@ structure, not text.
 Until 2026-10-04 the builder read past these marks and L1 had no
 paragraphs at all (broken tablet 39). Adding them added this one field
 to 686 verse files and changed nothing else; the hash moved from
-`156b1d27…` to `f743b34b…`.
+`156b1d27…` to `f743b34b…` (and since, with the inverted nuns, to
+`9e1abe4e…`).
+
+## Inverted Nuns
+
+Two verses carry `inverted_nuns`:
+
+```json
+"inverted_nuns": [ { "after_word": 7 } ]
+```
+
+The *nun hafukha* (׆) stands after Bamidbar 10:34 and after 10:36,
+marking off *Vayehi binsoa ha'aron* (10:35–36) above and below — "because
+it is a book of its own" (Shabbat 115b–116a). The WLC writes the sign
+after the verse's sof pasuq, before the setumah; like a paragraph break
+it is recorded by the word it follows, and it is not part of
+`text_hebrew`. Added 2026-10-04 (broken tablet 40); the hash moved from
+`f743b34b…` to `9e1abe4e…`.
+
+## Large and Small Letters
+
+A word with an enlarged letter carries `large_letters`. The WLC marks
+four in Torah, and L1 has those four: the vav of גחון (Vayikra 11:42),
+the nun of משפטן (Bamidbar 27:5), and the ayin and dalet of the Shema
+(Devarim 6:4).
+
+**L1 records no small letters, because its source marks none.** The WLC
+as the Open Scriptures Hebrew Bible carries it has no small-letter mark
+anywhere in the five books — not the aleph of ויקרא (Vayikra 1:1), not
+the he of בהבראם (Bereishit 2:4). The builder reads `x-small` and would
+record one if the source had it. The enlarged and reduced letters of the
+later Masoretic lists (the bet of בראשית among them) are a tradition of
+the scroll that this edition of the Leningrad Codex does not encode;
+adding them would mean engraving something the source does not say.
 
 ## Word Fields
 
diff --git a/l1/build_report.json b/l1/build_report.json
index 5a08e8fd..2474db8b 100644
--- a/l1/build_report.json
+++ b/l1/build_report.json
@@ -9,6 +9,7 @@
     "petuchah": 293,
     "setumah": 394
   },
+  "inverted_nuns": 2,
   "words_with_shoresh": 57487,
   "shoresh_provenance": {
     "main": 23750,
@@ -18,7 +19,7 @@
     "none": 3002,
     "unlisted": 0
   },
-  "sha256": "f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494",
+  "sha256": "9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086",
   "missing_teamim_refs": [],
   "words_without_lexicon_root": [
     {
diff --git a/l1/integrity.sha256 b/l1/integrity.sha256
index f62ce138..9fb75abd 100644
--- a/l1/integrity.sha256
+++ b/l1/integrity.sha256
@@ -1 +1 @@
-f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494
\ No newline at end of file
+9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086
\ No newline at end of file
diff --git a/l1/torah/bamidbar/10/34.json b/l1/torah/bamidbar/10/34.json
index 8492f465..2c4553e4 100644
--- a/l1/torah/bamidbar/10/34.json
+++ b/l1/torah/bamidbar/10/34.json
@@ -97,5 +97,10 @@
       "after_word": 7,
       "type": "setumah"
     }
+  ],
+  "inverted_nuns": [
+    {
+      "after_word": 7
+    }
   ]
 }
\ No newline at end of file
diff --git a/l1/torah/bamidbar/10/36.json b/l1/torah/bamidbar/10/36.json
index b61ebf63..433592fd 100644
--- a/l1/torah/bamidbar/10/36.json
+++ b/l1/torah/bamidbar/10/36.json
@@ -106,5 +106,10 @@
       "after_word": 7,
       "type": "setumah"
     }
+  ],
+  "inverted_nuns": [
+    {
+      "after_word": 7
+    }
   ]
 }
\ No newline at end of file
diff --git a/scripts/build_l1.py b/scripts/build_l1.py
index bec03565..3c6446c2 100644
--- a/scripts/build_l1.py
+++ b/scripts/build_l1.py
@@ -421,6 +421,7 @@ def parse_verse(verse_elem, book_osis: str, book_hebrew: str) -> dict:
     word_position = 0
     ketiv_qere_entries = []
     parashah_breaks = []
+    inverted_nuns = []
 
     children = list(verse_elem)
     i = 0
@@ -477,6 +478,16 @@ def parse_verse(verse_elem, book_osis: str, book_hebrew: str) -> dict:
                     "after_word": word_position,
                     "type": PARASHAH_BREAKS[seg_type],
                 })
+            elif seg_type == "x-reversednun":
+                # Nun hafukha (׆). Two in Torah, bracketing "Vayehi
+                # binsoa ha'aron" (Bamidbar 10:35–36): one after 10:34
+                # and one after 10:36. Shabbat 115b–116a: the passage is
+                # marked off above and below because it is a book of its
+                # own. The WLC writes the sign after the verse's sof
+                # pasuq; it is recorded, like a paragraph break, by the
+                # word it follows. Until tablet 40 the builder read past
+                # it.
+                inverted_nuns.append({"after_word": word_position})
 
         i += 1
 
@@ -507,6 +518,8 @@ def parse_verse(verse_elem, book_osis: str, book_hebrew: str) -> dict:
         verse_out["smikhut"] = smikhut
     if parashah_breaks:
         verse_out["parashah_breaks"] = parashah_breaks
+    if inverted_nuns:
+        verse_out["inverted_nuns"] = inverted_nuns
     return verse_out
 
 
@@ -730,6 +743,7 @@ def build_l1(morphhb_dir: str, output_dir: str, lexicon_dir: str):
                 total_verses += 1
                 for brk in verse_data.get("parashah_breaks", []):
                     parashah_counts[brk["type"]] += 1
+                parashah_counts["inverted_nun"] += len(verse_data.get("inverted_nuns", []))
 
                 # Count words and collect stats
                 for word in verse_data["words"]:
@@ -802,6 +816,7 @@ def build_l1(morphhb_dir: str, output_dir: str, lexicon_dir: str):
     print(f"Verses missing te'amim: {len(missing_teamim)}")
     print(f"Parashah breaks:      {parashah_counts['petuchah']} petuchot, "
           f"{parashah_counts['setumah']} setumot")
+    print(f"Inverted nuns:        {parashah_counts['inverted_nun']}")
     words_with_shoresh = sum(_SHORESH_STATS[k] for k in ("main", "sub", "single"))
     rootless = sorted(_ROOTLESS.items(), key=lambda kv: (-kv[1], kv[0]))
     print(f"Words with a shoresh: {words_with_shoresh}")
@@ -840,6 +855,7 @@ def build_l1(morphhb_dir: str, output_dir: str, lexicon_dir: str):
             "petuchah": parashah_counts["petuchah"],
             "setumah": parashah_counts["setumah"],
         },
+        "inverted_nuns": parashah_counts["inverted_nun"],
         "words_with_shoresh": words_with_shoresh,
         "shoresh_provenance": {k: _SHORESH_STATS[k] for k in
                                ("main", "sub", "single", "particle", "none", "unlisted")},
diff --git a/shoresh-engine/src/Luchot/Firmware.hs b/shoresh-engine/src/Luchot/Firmware.hs
index 2d0826f7..a8c167bb 100644
--- a/shoresh-engine/src/Luchot/Firmware.hs
+++ b/shoresh-engine/src/Luchot/Firmware.hs
@@ -30,7 +30,7 @@ import Data.Text (Text)
 -- From build_report.json and verified during L1 build.
 -- If the hash does not match, the specification is compromised.
 l1ExpectedHash :: Text
-l1ExpectedHash = "f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+l1ExpectedHash = "9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 
 -- | Torah book count. Five books. Not four, not six.
 torahBookCount :: Int
diff --git a/shoresh-engine/test/Spec.hs b/shoresh-engine/test/Spec.hs
index d9423a60..d0316427 100644
--- a/shoresh-engine/test/Spec.hs
+++ b/shoresh-engine/test/Spec.hs
@@ -314,7 +314,7 @@ firmwareSpec :: Spec
 firmwareSpec = describe "Firmware" $ do
   it "has correct L1 hash" $
     l1ExpectedHash `shouldBe`
-      "f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+      "9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 
   it "has correct counts" $ do
     torahBookCount `shouldBe` 5
diff --git a/specs/shorashim_terumah.md b/specs/shorashim_terumah.md
index 11d375ce..26ee4a77 100644
--- a/specs/shorashim_terumah.md
+++ b/specs/shorashim_terumah.md
@@ -121,6 +121,8 @@ Source: Westminster Leningrad Codex (WLC) 4.20, as carried by the Open Scripture
 | Full Torah text — voweled (nikkud) | ✅ | WLC via OSHB morphhb. |
 | Cantillation marks (raw Unicode codepoints U+0591–U+05AF) | ✅ | Present in WLC data. Stored as raw marks on the text, NOT pre-parsed into trees. The system reads them in the act of reading. |
 | Petuchah/Stumah paragraph markers | ✅ | In L1: 293 petuchot and 394 setumot, read from the WLC's `פ` / `ס` marks (OSIS `<seg type="x-pe">`, `x-samekh`) and stored per verse as `parashah_breaks`. |
+| Inverted nuns (nun hafukha) | ✅ | In L1: the two signs after Bamidbar 10:34 and 10:36, from the WLC's own marks, stored per verse as `inverted_nuns`. |
+| Large and small letters | ⚠️ | The WLC source marks four large letters in Torah (in L1 as `large_letters`) and NO small letters. The fuller Masoretic lists of enlarged and reduced letters are not in the source and not in L1. |
 | Ketiv/qere annotations | ✅ | WLC annotations: 67 pairs in 64 verses. Stored per verse as `{word_position, ketiv, ketiv_consonants, qere, qere_consonants}` — both readings preserved, never collapsed. |
 | Gematria (root → numerical value) | ✅ | Not stored in L1: computed from the consonants when read. Letter values fixed. No equivalence index is built. |
 | Smikhut compound index (attested construct-state pairs) | ✅ | Built: 9,203 pairs in 4,106 verses, extracted by `scripts/build_l1.py` from the morphology where a word is marked construct. |
diff --git a/specs/shorashim_vayakhel.md b/specs/shorashim_vayakhel.md
index 542389f2..df87c0d5 100644
--- a/specs/shorashim_vayakhel.md
+++ b/specs/shorashim_vayakhel.md
@@ -45,14 +45,15 @@ The specification that sits inside the aron must be specified with the same prec
 | **Source — roots** | `openscriptures/HebrewLexicon` at commit `21c9add` (`LexicalIndex.xml`, `AugIndex.xml`): each lemma filed under its Brown-Driver-Briggs root. BDB and Strong's public domain; the index CC BY 4.0 |
 | **Builder** | `scripts/build_l1.py`. Deterministic: the same two checkouts give the same bytes and the same hash. Both commits are recorded in `l1/build_report.json` |
 | **Extent** | 5 books, 187 chapters, 5,853 verses (WLC/BHS versification; the traditional count is 5,845), 79,982 words |
-| **Format** | One JSON file per verse, `l1/torah/{book}/{chapter}/{verse}.json`: `{ "ref": "Bereishit 1:1", "book": "bereishit", "chapter": 1, "verse": 1, "text_hebrew": "...", "words": [...] }`, with `"ketiv_qere": [...]`, `"smikhut": [...]` and `"parashah_breaks": [...]` where the verse has them |
-| **Word** | `{ "position": 1, "text": "בְּרֵאשִׁ֖ית", "consonants": "בראשית", "lemma": "7225", "morph": "HR/Ncfsa", "teamim": ["tipcha"], "prefix": "ב", "shoresh": "ראש" }`. `prefix` and `shoresh` are present only where the word has them; `large_letters` marks an enlarged letter (4 words) |
+| **Format** | One JSON file per verse, `l1/torah/{book}/{chapter}/{verse}.json`: `{ "ref": "Bereishit 1:1", "book": "bereishit", "chapter": 1, "verse": 1, "text_hebrew": "...", "words": [...] }`, with `"ketiv_qere": [...]`, `"smikhut": [...]`, `"parashah_breaks": [...]` and `"inverted_nuns": [...]` where the verse has them |
+| **Word** | `{ "position": 1, "text": "בְּרֵאשִׁ֖ית", "consonants": "בראשית", "lemma": "7225", "morph": "HR/Ncfsa", "teamim": ["tipcha"], "prefix": "ב", "shoresh": "ראש" }`. `prefix` and `shoresh` are present only where the word has them; `large_letters` marks an enlarged letter (4 words — all the source marks). No small letters are recorded: the source marks none in the five books |
 | **Integrity** | SHA-256 over every file of the L1 tree (path, then bytes), recorded at build time in `integrity.sha256`. The pipeline re-reads and re-hashes the tree at startup and every ten minutes |
 | **Cantillation marks** | Kept on the text as the Unicode codepoints the WLC carries (U+0591–U+05AF), and named per word in `teamim`. The parse tree is NOT pre-computed: the reader builds it from the marks |
 | **Ketiv/qere** | From the WLC's ketiv/qere annotations: 67 pairs in 64 verses, stored as `{ "word_position": 6, "ketiv": "...", "ketiv_consonants": "...", "qere": "...", "qere_consonants": "..." }` |
 | **Shoresh** | Read from the lexicon by the word's lemma — never derived from the consonants. 57,487 words carry one; 1,378 distinct roots. Particles and pronouns carry none, by rule; names the lexicon gives no etymology carry none |
 | **Gematria** | Not stored. Computed from the consonants when read (engine and pipeline) |
 | **Petuchah / setumah** | From the WLC's own `פ` / `ס` marks: 293 petuchot and 394 setumot in 686 verses, stored as `"parashah_breaks": [{ "after_word": 13, "type": "petuchah" }]`. The break is recorded by the word it follows; four fall inside a verse. These are the Leningrad Codex's divisions |
+| **Inverted nuns** | The two *nun hafukha* signs, after Bamidbar 10:34 and after 10:36, from the WLC's own marks: `"inverted_nuns": [{ "after_word": 7 }]` |
 | **Smikhut compounds** | From the morphology, where a word is marked construct: 9,203 attested pairs in 4,106 verses, stored as `{ "position": 5, "construct": "...", "construct_consonants": "...", "absolute": "...", "absolute_consonants": "..." }` |
 
 **Verification procedure:**
diff --git a/vav_pipeline/lib/vav_pipeline/doorpost.ex b/vav_pipeline/lib/vav_pipeline/doorpost.ex
index 7bc6fd13..d45bb9c1 100644
--- a/vav_pipeline/lib/vav_pipeline/doorpost.ex
+++ b/vav_pipeline/lib/vav_pipeline/doorpost.ex
@@ -175,7 +175,7 @@ defmodule VavPipeline.Doorpost do
   # changes; the update is a deliberate operator commit. A bare
   # hash-update commit (without a corresponding spec change in the
   # same commit) is itself a drift signal.
-  @expected_hash "b62311e7541277f3881f074e2e884b742f6f85bd2de8b4890f44e978d06a3f21"
+  @expected_hash "3bc5dd6aa4d76eeb8c0bc2ee4d7d33b7f35bc1dc8e03dc76697acc33eec8e245"
 
   @doc "Return the list of files that constitute the category-grade specification."
   @spec files() :: [String.t()]
diff --git a/vav_pipeline/lib/vav_pipeline/l1_reader.ex b/vav_pipeline/lib/vav_pipeline/l1_reader.ex
index 2f851585..248710f4 100644
--- a/vav_pipeline/lib/vav_pipeline/l1_reader.ex
+++ b/vav_pipeline/lib/vav_pipeline/l1_reader.ex
@@ -17,7 +17,7 @@ defmodule VavPipeline.L1Reader do
   # Updated after build_l1.py fix: enlarged letters (Shema's ayin/dalet, gachon's
   # vav, mishpatan's nun), ketiv/qere pairs, and smikhut construct pairs are now
   # preserved in L1 rather than silently dropped.
-  @l1_expected_hash "f743b34b83617c328ec190e78ea1402470938e460b9701aec465c3f1613f1494"
+  @l1_expected_hash "9e1abe4edabac825d8a7037606ff247c3e8f8167650f04553f46c34e4f3c3086"
 
   @type letter_mark :: %{letter: String.t(), offset: non_neg_integer()}
 
@@ -64,7 +64,8 @@ defmodule VavPipeline.L1Reader do
           words: [word()],
           ketiv_qere: [ketiv_qere_entry()],
           smikhut: [smikhut_pair()],
-          parashah_breaks: [parashah_break()]
+          parashah_breaks: [parashah_break()],
+          inverted_nuns: [%{after_word: pos_integer()}]
         }
 
   @doc """
@@ -386,7 +387,13 @@ defmodule VavPipeline.L1Reader do
       words: Enum.map(data["words"] || [], &normalize_word/1),
       ketiv_qere: Enum.map(data["ketiv_qere"] || [], &normalize_ketiv_qere/1),
       smikhut: Enum.map(data["smikhut"] || [], &normalize_smikhut/1),
-      parashah_breaks: Enum.flat_map(data["parashah_breaks"] || [], &normalize_parashah_break/1)
+      parashah_breaks: Enum.flat_map(data["parashah_breaks"] || [], &normalize_parashah_break/1),
+      # Nun hafukha (׆) — recorded, like a paragraph break, by the word it
+      # follows. Two in Torah: after Bamidbar 10:34 and after 10:36.
+      inverted_nuns:
+        for %{"after_word" => n} when is_integer(n) <- data["inverted_nuns"] || [] do
+          %{after_word: n}
+        end
     }
   end
 
diff --git a/vav_pipeline/test/shema_voice_trajectory_test.exs b/vav_pipeline/test/shema_voice_trajectory_test.exs
index 66d76d8e..f02c8925 100644
--- a/vav_pipeline/test/shema_voice_trajectory_test.exs
+++ b/vav_pipeline/test/shema_voice_trajectory_test.exs
@@ -20,7 +20,7 @@ defmodule VavPipeline.ShemaVoiceTrajectoryTest do
   stabilizing at one pole — which is the algorithmic shape of "you have
   forgotten the One who took you out of Egypt".
 
-  Anchoring verses (text from L1, hash f743b34b…):
+  Anchoring verses (text from L1, hash 9e1abe4e…):
     Devarim 6:5  וְאָהַבְתָּ … אֱלֹהֶיךָ … לְבָבְךָ — uniformly 2ms
     Devarim 11:13 וְהָיָה אִם־שָׁמֹעַ תִּשְׁמְעוּ … אֱלֹהֵיכֶם — 2mp
     Devarim 11:14 אַרְצְכֶם (2mp) → וְאָסַפְתָּ דְגָנֶךָ (2ms)
diff --git a/vav_pipeline/test/tablet40_inverted_nuns_test.exs b/vav_pipeline/test/tablet40_inverted_nuns_test.exs
new file mode 100644
index 00000000..03849201
--- /dev/null
+++ b/vav_pipeline/test/tablet40_inverted_nuns_test.exs
@@ -0,0 +1,55 @@
+defmodule VavPipeline.Tablet40InvertedNunsTest do
+  @moduledoc """
+  Tablet 40 (2026-10-04) — the inverted nuns.
+
+  *Vayehi binsoa ha'aron* (Bamidbar 10:35–36) is marked off in the scroll
+  by a nun hafukha before it and after it. The WLC carries both signs; the
+  builder read past them. They are now engraved on the verse each one
+  follows — 10:34 and 10:36 — and the reader carries them.
+  """
+
+  use ExUnit.Case, async: true
+
+  alias VavPipeline.L1Reader
+
+  test "an inverted nun follows Bamidbar 10:34 and another follows 10:36" do
+    for verse <- [34, 36] do
+      {:ok, v} = L1Reader.read_verse("bamidbar", 10, verse)
+      assert v.inverted_nuns == [%{after_word: length(v.words)}], "Bamidbar 10:#{verse}"
+    end
+  end
+
+  test "the two verses between them carry none — the signs bracket the passage" do
+    {:ok, v35} = L1Reader.read_verse("bamidbar", 10, 35)
+    assert v35.inverted_nuns == []
+    assert hd(v35.words).consonants == "ויהי"
+    assert Enum.at(v35.words, 1).consonants == "בנסע"
+  end
+
+  test "each nun stands with a setumah, and neither is in the verse's text" do
+    for verse <- [34, 36] do
+      {:ok, v} = L1Reader.read_verse("bamidbar", 10, verse)
+      assert v.parashah_breaks == [%{after_word: length(v.words), type: :setumah}]
+      refute v.text_hebrew =~ "׆"
+    end
+  end
+
+  test "there are exactly two in the whole scroll" do
+    index = L1Reader.l1_root() |> Path.join("torah/index.json") |> File.read!() |> Jason.decode!()
+
+    refs =
+      for book <- index["books"],
+          {chapter, verses} <- book["chapter_verse_counts"],
+          verse <- 1..verses,
+          {:ok, v} = L1Reader.read_verse(book["book"], String.to_integer(chapter), verse),
+          v.inverted_nuns != [],
+          do: v.ref
+
+    assert Enum.sort(refs) == ["Bamidbar 10:34", "Bamidbar 10:36"]
+  end
+
+  test "every other verse carries an empty list, not a missing field" do
+    {:ok, v} = L1Reader.read_verse("bereishit", 1, 1)
+    assert v.inverted_nuns == []
+  end
+end