Topic 24 "Where Earth's water is" (salty, fresh and frozen): your read (2 October 2026)

The blind reads. Two readers answered the first bundle cold and agreed every key (127 items each). Their flags: a stem whose answer the lesson title steered ("Rain is fresh water"); a drip that could be precipitation or collection; "before it fell as rain" that strictly means "in a cloud"; term items at the end of L01's quiz with "water vapor" used before L01 teaches it; "glacier" never defined; "iceberg" untaught; two lesson drawings relettered as bank items; British words ("washing-up water", "taps"); two page give-aways in fluency quizzes. All applied in v2. Reader C then read the v2 bundle whole (130 items) and agreed every key; two fixes and four checks, all applied in v3 (the polish table in the build record).

The guidance, questions and accuracy review. The two biggest findings: L02 re-taught Grade 3 Topic 28's reservoir lesson almost scene for scene and dropped the coined word "reservoir" (now one recalled line and the word; the hose-on-the-lawn scene became the town turning off its fountains; L02 is 294 words); and the bank's keys were test-wise bait (26 of 49 the single longest option, 8 the only two-clause option; every key is now one clause and never the single longest; the vault's key-tell checker reads 0 of 122). Worth-fixing rows: a false "so" in L02's rule and "living things" against the sea fish (now "People, animals and plants on land need fresh water…"); the model's only solid standing for cold air (one line names the real solid water); two rule cards not word for word; "Farm A / B"; a lake that freezes; no bank recall of L01's coined words (PP-12 and PP-41 added); "shows he is wrong" stems made positive; D under-used in the bank's keys.

The figure review. The hundred-squares grid counts true on the render (97 blue, 2 white, 1 green). The model's chips floated: LIQUID beside dots and GAS beside dots read as two identical regions with opposite names, and the falling drop sat on a closed cup. The model drawer was rewritten: open cup, the drop above its rim, every chip on its thing or leadered to it. The ocean-to-rain landscape had its lake chip in the soil and a grey vapor arrow rising straight at the Sun (the "Sun pulls the water" reading the language rule forbids); now the lake chip sits on the lake and Topic 23's yellow Sun-to-water arrow carries the energy. The fresh-water zoom clipped its text; a caption said "real"; four unplaced figures deleted and the globe photo placed.

57 items: 22 easy, 18 medium, 17 hard, sorted easy to hard with the atoms rising inside each tier. Per atom: EAR4-058 14 · 059 9 · 060 10 · 061 11 · 062 13. Forms: 50 four-option MCQs, 6 typed (ice cap, the ocean twice, salt, gas, liquid), 1 multi-select (the liquid parts of the tub model). MCQ keys A 10 · B 18 · C 12 · D 10. Every item is a fresh scene (astronauts, a harbor, a cactus farm, a lighthouse keeper, a horse ranch, fog over a hilly city, salt pans, an island hotel, a dairy farm and a vegetable farm, a prairie storm, a sailboat, a rain barrel, a canal, and a tub model that is never the lesson's bowl). Thirteen question types, one exemplar each, on the type sheet, with the draw constraints (pairs never served together because one states what the other asks). The hard tier types every coined word the grade can spell and recalls the two-word kinds by MCQ. "Hard because" lines: produce the word or the place from the rule; predict a changed model; read a drawn item; order a drop's journey; chain the 9-and-1 shares.

Skill: course-build → write-lesson (L01 / L03 / L04 facts true of every case; L02 an explanation with its claims; L05 the routine capstone) → write-example-sequence, design-check-sequence, sentence-atomise, write-explanation, write-mcqs (quizzes, attached questions, practice, the PP100 bank and its type sheet); write-video-script (intro, five lesson scripts, summary) @ 3a2c39d

Grade 4 Topic 24 — Where Earth's water is — build record (v1, 2 October 2026)

What this is, in plain words

Five lessons that answer Levi's question at the globe: Earth is nearly all blue, so will we never run short of water to drink? The children count Earth's water as 100 squares (97 salt water in the ocean, 2 frozen, 1 fresh water we can reach, nearly all of it under the ground), see why the little fresh water matters when the rain stops, learn that most of the water in the air evaporated from the ocean, find out why the rain is fresh anyway (evaporation leaves the salt behind) and that rivers carry the water back so the ocean is connected to all of it, and then build the water cycle on a windowsill and label where the water is a solid, a liquid or a gas. Everything is written to the Grade 4 standing brief with no blueprint; the structural decisions a blueprint would have made are listed below.

Counts. 5 lessons · 5 scripts (150, 146, 150, 149, 148 spoken words) + intro (118 words, 2 question marks) + summary (174 words, opener exact) · 5 attached questions + the intro's poll + 3 summary questions (key string in video order A A B B A B C C, A 3 of 8) · 49 quiz items (12 + 9 + 8 + 9 + 11) · 15 practice items (13 MCQ, 1 written answer, 1 typed; three Grade 3 retrievals first; the three term recalls last) · PP100 bank 55 items (20 easy · 19 medium · 16 hard; per atom 11 · 10 · 10 · 11 · 13; 5 typed, 1 multi-select, 2 drawn; MCQ keys A 11 · B 17 · C 12 · D 9) · 48 figures (35 svg, 9 photo placeholders, 4 cartoon placeholders; 13 image briefs). Articles 295 / 337 / 298 / 322 / 393 words (see "Article length").

Decisions made without a blueprint (one line each, with the reason)

• Five lessons, one atom each, in the register's order 058 → 059 → 060 → 061 → 062: every prerequisite edge points backward (059 needs 058; 060 needs 056 and 058; 061 needs 060; 062 needs 055, 057, 060), and the capstone closes the topic as the architecture row says.
• One running cast and one running question: Levi's "we will never run short of water to drink" at the globe opens the intro and L01, and each later lesson re-establishes its own scene (a dry July; a rain cloud over the town; a splash at the edge of the sea; the windowsill), because every video must stand alone (James, 2 Oct 2026).
• The hundred-squares grid is the teaching figure for EAR4-058, drawn by code (exact), with the rounding 97 / 2 / 1 (see "Numbers and rounding"); the headache figure is the globe photo, never the grid, because the grid's counts are the answer's datum.
• Three coins in L01 — "salt water", "fresh water", "ice cap" — and no more: "glacier" is Topic 20's word, recapped once in its own sentence; "water vapor" is Grade 3's; "groundwater" stays a Grade 5 word ("water under the ground, in the tiny spaces between grains of soil and rock").
• L02 is an explanation (type c), not a fact run: the atom is an argument (why the small share matters), so the article runs three chains (living things need fresh water and salt water damages them; the town's lake runs short when used faster than rain replaces it — Grade 3 Topic 28's sentence word for word; only 1 square of 100 is reachable) and the quiz is what / why / claims / transfer.
• One boundary line in L02 — some seaside towns do take the salt out of sea water, with huge machines and a lot of energy: a child who has just heard "the ocean is no use" asks it at once; the line answers without teaching desalination.
• L03 teaches "most of the water in the air comes from the ocean" with the ten cups (about 9 of 10) and the reason "because the ocean is where most of the water is", recalling L01's 97 squares and Topic 23's Sun sentence word for word; the wind carrying clouds across the land is one everyday line, not a taught mechanism.
• L04 kills the salty-rain belief on a dish of sea water in the sun (Grade 3 Topic 21's strip convention, a new drawer), recalls Topic 14's "A dissolved material is not gone" word for word, then draws the leg Topic 23 never drew as a landscape (ocean → vapor → cloud → mountain lake) and imports Topic 23's where-the-water-goes cross-section read-only for the river back to the sea.
• L05 is a routine (type b): the class's model is the worked example on the page, Levi's model at home (a pan, a mug, plastic wrap, an ice cube) carries the scaffolded items, and the drawn model with letters plus two BEFORE cards carry the bare items — scaffolded items never sit on the worked case.
• The model's ice is "the cold spot, like the cold air high up", and its water is labelled solid: the register asks for a labelled model marking solid / liquid / gas, and the ice's job in a real model is to make vapor condense; calling the ice "the ice caps" would teach that clouds form on glaciers.
• The cup's water is stated to be fresh, never tasted: Grade 3's rule "In science, we don't taste things to observe them" (Topic 16 ruling) — L04 has already taught why the salt stays behind.
• Videos: every lesson is "video and article" — every atom here has a picture a child needs to see move (squares filling, cups lighting, dots rising, drops forming).
• No STRAND_INTRO: Topic 17 opened Earth and Space.

Questions for James (one line each; the build did not wait)

1. Is 97 / 2 / 1 the right rounding for the hundred squares? USGS gives 96.5 % ocean, 1.74 % ice caps and glaciers, 1.69 % under the ground (about half of that salty), 0.013 % lakes, 0.0002 % rivers. Recommendation: keep 97 / 2 / 1 with "nearly all of the 1 under the ground; rivers and lakes a tiny slice" — three whole squares a child can count. Built so.
2. Is "about 9 of every 10 cups" (the ocean's share of evaporation, about 86 %) close enough? Recommendation: keep; "nearly 9" buys nothing at Grade 4. Built so.
3. Should the desalination line stay in L02? Recommendation: keep the one line; the question is the first a bright child asks, and the line names the cost without a term.
4. Should the model's ice be "the cold spot" (built) or "the ice caps"? Recommendation: the cold spot, as built; the ice is still labelled solid, so the register's solid / liquid / gas labelling is met.
5. L05's article runs 393 words (the five numbered build steps, what happens, the states run). Recommendation: keep it whole — a routine split by a check loses its order; a subtitle or figure sits every ≤ 6 lines. Built so.
6. Topic 23's loop figure (imported read-only) stacks its three top chips (THE SUN'S LIGHT / CARRIES ENERGY / CONDENSATION) tightly on the L03 page. Not mine to change; noted for Topic 23's author.

What each lesson teaches

#### design-check-sequence records

`

Skill: design-check-sequence @ 3a2c39d

Part: L01 (one part)

Type: (a) — the student places a water in salt / fresh, or names where most of Earth's water (or fresh water) sits; one step each.

Plan: 12 items: 1 Grade 3 recall (ice and water are the same stuff, yes / no) → 2 fresh / salt sorts (2 options) → where most fresh water is (4) → how many squares reachable (3, ordered) → 4 claims bare yes / no (most is drinkable — no; rivers hold most — no; ice caps are not water — no; most fresh water is frozen — yes) → the grid read (3 letters) → 2 term recognitions last (fresh water; ice cap). Two-option string A A B B B B A; three-option A C B C; four-option B.

Belongs in practice/test instead: the typed "ice cap" (PR-15); the pick-the-one-of-four forms (bank types 1–2).

Part: L02

Type: (c) — the payoff is a because-clause: why the small share of usable fresh water matters.

Plan: 9 items: Grade 3 recall (a plant with no water; 3) → Grade 3 rule recall (runs short when…; 3) → the what (the lake gets lower; 3) → the why (uses faster than rain replaces; 3 reasons) → claim (seaside town can never run short — no) → claim (corn needs fresh water — yes) → the reason for a camping group (3) → transfer (a sheep farm's spring; 3) → what shorter showers do (3). Two-option B A; three-option B A C B C B C. The why sits on the SAME case as the what (the town's lake), stating the scene and the observation, never the what's key sentence (the MS Chemistry form; recorded).

Belongs in practice/test instead: the comparative "which pond runs short first" (PP-22); the island-hotel claim (PP-21).

Part: L03

Type: (a) — one-step: where did this water evaporate from; the reason is one fixed sentence.

Plan: 8 items: Topic 23 / Grade 3 recall (what carries energy to the water; 3) → the 9 of 10 number (3, ordered) → two source items on fresh scenes (snow on a mountain; which place sends the most; 3 each) → the why (3 reasons) → 3 claims bare yes / no (only the river — no; mostly the ocean — yes; only from right under the cloud — no). Two-option B A B; three-option A C B A C. Eight items: a one-step lesson gets as many honest items as exist.

Belongs in practice/test instead: the thought experiment (PP-30); the year-after-year chain (PP-41); the typed place (PP-40).

Part: L04

Type: (a) with one what / why pair — one step: what is left, which water the rain is, where the water was or goes.

Plan: 9 items: Topic 14 recall (dissolved is not gone; yes / no) → Grade 3 recall (a puddle left open; 3) → the what (what is left in the pan; 3) → the why (why rain is fresh; 3 reasons; same case as the lesson's dish, fresh pan) → claim (coast rain salty — no) → trace back (mountain lake drop; 3) → trace forward (a city river; 3) → claim (river water was once in the ocean — yes) → term recognition for salt water last (3). Two-option A B A; three-option A B C B C C.

Belongs in practice/test instead: the drop's story in order (PP-44); the typed salt (PP-42); the landscape letter read (PP-43).

Part: L05

Type: (b) — a multi-step routine: build the model (five steps), then label parts, states and steps.

Plan: 11 items: worked example = the class's model on the page (shown, not asked). Scaffolded, on Levi's model at home (pan, mug, wrap, ice cube): which part is the ocean (4) → the vapor's state (3) → the drops' step (4) → the drip's step (4) → fresh or salt in the mug (2) → the solid part (3) → what the ice stands for (3). Bare, on a third case: the drawn lettered model (which letter is a gas; 4) → two predictions on BEFORE cards (dark cupboard; nothing cold on top; 2 and 3 options, ordered) → claim (the cup's water is salty — no). Four-option A B C A; three-option B C B A; two-option A A B.

Belongs in practice/test instead: the tub model (bank type 10–11); the typed states and the typed real thing (PP-48–50); the multi-select (PP-54).

`

The key strings

• Attached questions, video order (L01 … L05, summary ×3): A A B B A B C C (A 3 of 8, B 3 of 8, C 2 of 8).
• Quizzes: L01 two-option A A B B B B A, three-option A C B C, four-option B · L02 two-option B A, three-option B A C B C B C · L03 two-option B A B, three-option A C B A C · L04 two-option A B A, three-option A B C B C C · L05 four-option A B C A, three-option B C B A, two-option A A B. No four keys in a row alternate; no position above 60 % in any lesson's same-size set (checker S7 clean).
• Practice MCQs (13): A B C A A C B B A C A C B.
• Bank MCQs (49): A 11 · B 17 · C 12 · D 9; fixed sets (PLACE4, SOURCE4, STEP4) keep their order and are counted where they land.

The 2 October rules, applied

Numbers and rounding (named simplifications)

• 97 / 2 / 1 of 100 squares. USGS: ocean 96.5 %; ice caps, glaciers and permanent snow 1.74 %; groundwater 1.69 % (about 0.76 % fresh); lakes 0.013 %; rivers 0.0002 %; the air 0.001 %. Rounded to three whole squares; the fresh_zoom bar shows "under the ground: nearly all of the square" and "rivers and lakes: a tiny slice". The salty half of the groundwater is folded into "under the ground" without comment (a Grade 5 refinement).
• About 9 of every 10 cups of water that go into the air come from the ocean. The ocean's share of global evaporation is about 86 %; plants' transpiration is folded into "wet ground".
• The ocean covers most of Earth's surface (71 %) — said in words, no number, because a second set of numbers beside the hundred squares would confuse the two shares.
• "A model is a small copy of the real thing that works the same way." — the plain meaning, no term coined.

Misconceptions confronted (the vault catalogue is physics-only and silent on water; sources named)

Checker: 0 FAIL, 41 WARN (python3 build_tools_check_topic.py grade4/topic24_build/content/G4T24.py)

• S8 (15) script beats of three to five sentences — kept: each beat is one screen state (the Topic 20 shape); every script is ≤ 150 spoken words (S25 clean).
• S36 (26 words), each answered: salty ×18 (an everyday taste word — fine); globe ×10 (the classroom object; fine); condensed / condense ×11 (Grade 3 Topic 21's coined word in another tense — fine); prairie ×4, reef, coral, tide, pelican, iceberg (US words a Grade 4 child meets; each in one bank scene — fine); droops / droop ×5 (Grade 3 Topic 33's own word for the tomato plant — fine); pilot, hiker, scientist (everyday roles — fine); grid ×2 (the figure's title says "100 squares"; "grid" appears in the two item stems with the drawing in view — fine); collects (the step word's verb, Topic 23 — fine); uphill ×2 (Topic 20's word — fine); usual, replacing / replace, shallow, chunk, spray, heavier, inland (everyday — fine; inland survives in one option after the stem took "far from the coast"). Swapped before this run: sliver → a tiny slice (13 places), enlarged → made bigger, thirstier → even more thirsty, filter → take the salt out, sag / sags → the low point.
• Clean: S1–S7, S9–S17, S19–S35, S37–S39. S30 typed=6 fail=0 warn=0; S35 typed_long=0; S37 strand=0; S38 fragments=0; S39 hits=0; S29 units 0.
• python3 grade4/check_g4_articles.py … → 0 FAIL: scripts 150 / 146 / 150 / 149 / 148; intro 118; summary 174, opener exact.

#### Article length (brief: 170–300, WARN above 300 acceptable with a subtitle or figure every ≤ 8 lines and a reason)

L01 295 · L02 337 (three chains, each under its own subtitle with a table or photo; the salt-out boundary has its own subtitle) · L03 298 · L04 322 (the dish strip, the landscape and Topic 23's cross-section each sit under a subtitle; no run over 5 idea lines) · L05 393 (the five numbered build steps with the build cards after step 3 and the parts figure after step 5, then what happens, then the states run; no run over 6 idea lines; the routine is kept whole — question 5 to James). S16 clean on every lesson.

Bank freshness, re-performed

Scene nouns of every lesson, quiz, attached and summary item, grepped against the bank and the practice set (\b<noun>\b, case-insensitive): globe, desk, lawn(s), hose, yard, pipes, beach, wave, mouth, dish, crust, bowl, cup, wrap, bag, windowsill, jug, pan, mug, cupboard, well, ship, hiker, stream, tomato, pot, sheep, spring, camping, tent, shower(s), snow, pilot, Pacific, cornfield, bottle, glass, spoon, puddle — 0 bank hits on every one after the pass (re-scened: the ocean-wave spray → "spray blown off the top of the sea"; a mountain stream → "a clear forest river"; the seaside city → an island hotel; "short showers" → "turning off the tap while brushing teeth"; the mountain-lake claim → a rain barrel; the city river → a canal; the puddle on a rock → "a thin layer of sea water"; the attached L03 question's farm pond → a forest cloud; the summary's tray of ice → a block of ice; PR-6's village stream → an apple orchard's pump; PR-9's hiker → a deer at a forest river). Content nouns shared by design and not scenes: ocean, sea, water, rain, cloud, river, lake, town, farm, pond, mountain, ice, salt (the atoms are about these). The L05 model's nouns (bowl, cup, wrap, bag) appear 0 times in the bank: the bank's model is a tub, a jar, a glass lid and a tray of ice cubes.

Blind-bundle drawing lines: grep "\[Drawing:" BLIND_BUNDLE_G4T24.md | grep -ci "salt\|fresh\|most\|ocean\|frozen\|drink" → 0. The eight drawing lines list letters, colours, counts and positions only ("97 blue squares with chip P on them; 2 white squares with a sparkle…; 1 green square…" — the counts are data; "a blue block of water at the left under a yellow sun, chip N on it; tiny blue dots rising…"). No verdict word.

Figures (48; author's review-figures self-check on every render, then every built lesson page by screenshot)

Depiction register: content/figures.py docstring. Renders opened: three contact sheets of all new svg figures (grid, grid with letters, fresh_zoom, ten_cups, ocean_rain_labels and its letter form, salt_dish_strip, model parts / states / steps / after / start / two BEFORE cards / two letter forms, build_steps, five tables, five rule cards, the two Topic 23 imports), then the five article pages and the L05 quiz page at 900 px. Fixes from the renders: the ten-cups chip ran off the frame (frame widened, chip moved left); WATER VAPOR and RAIN chips overlapped the dots and the rain lines (moved); the mountain lake sat on a slope (moved into the valley); the cloud arrow's head was hidden under the dark cloud (shortened); CONDENSATION and PRECIPITATION chips overlapped the cup (moved right); build card 3's model was shifted off-centre (recentred). Everything else read correctly at a glance: labels point at the thing they name, every verdict chip names its question (the strip's MORNING / EVENING labels; SALT LEFT BEHIND), no item figure hands over its answer. check_svg_ids.py on the module: ids found 0, PASS (the module writes no defs at all; an assert in the module fails the import if any id ever appears).

Imported read-only from Topic 23: cycle_with_sun (L03) and where_labels (L04); nothing in Topic 23 was edited and no variant of its loop was drawn (the ocean-to-rain landscape draws the leg Topic 23 never drew). Topic 23's lesson file was NOT on disk when this topic was written — the recall lines ("The Sun's light carries energy to the water on Earth. The water warms up and evaporates."; "Fallen water gathers in streams, rivers, lakes, the sea and the ground."; the four step names) are taken from its figure module's rule cards and MUST be checked against grade4/topic23_build/content/G4T23.py lexicon when it lands (owed, below).

Sentence-atomise self-check (per lesson; the mechanical twin S1 / S2 / S38 is clean)

Judgement calls, recorded

• "the sea" against "the ocean": Topic 23 says "the sea"; this topic's atoms say "the ocean". One bridging line in L01 ("The sea you see from a beach is part of one huge ocean.") and "the ocean" everywhere after; the typed accept lists take both.
• The L02 why item sits on the same case as the what (the town's lake), stating only the scene and the observation — the MS Chemistry form; the fresh-case form is the AP Biology form. Recorded, not chosen silently.
• The bank keeps "lake", "pond", "farm", "town" as content nouns even though the lessons use them: the atoms are about where water sits and travels, so a lake in a bank item is the subject, not a borrowed scene; every bank scene's distinguishing noun (tub, jar, canal, prairie, golf course, rain barrel, kettle, lighthouse, hotel on an island) is new.
• UNBAN data row for G4T24 L03 ("energy") added to build_tools_video_batch.py — a data row only, as the Grade 4 chat agreed; Topic 23 will need the same row for its L04 (the Sun's energy drives the cycle) — flagged to the coordinator.
• The intro's one standing check_brief hit (closing: no bullets) is the R79-not-applicable-to-intros exception every topic carries; accepted, written in READY.md.
• The grid's fourth letter (R, the whole grid) exists only so the bank's grid item has four options without padding; the quiz's grid item has three letters and three options.
• No vault write-back this turn: no feedback from James was applied; one candidate note for the coordinator — the batch builder's banned-word list bans "energy" for every Grade 4 Earth topic that speaks Topic 23's Sun sentence, so each needs a data row.

Owed

• Blind review (bundle cut: blind/2026-10-02/BLIND_BUNDLE_G4T24.md, 119 MCQ + 6 typed + 1 FRQ + 1 multi; keys sequestered) — a non-author reads it; review-questions and review-figures passes by a non-author before James.
• The recall lines against Topic 23's lexicon when G4T23.py lands (see Figures).
• Videos: handoff/G4T24/videos/batch_G4T24.json + READY.md built and checked (6 of 7 clean, the intro's standing hit) — NOT sent; James's word is needed for Grade 4's first batch.
• The picture pass (13 placeholders: globe_class_cartoon, hose_cartoon, beach_cartoon, build_cartoon; globe, ocean, glacier, river, well, dry reservoir, rain over a mountain lake, town in the rain, salt dish, drops under the cover) through David's pack — after James's sign-off on the questions above.
• Nothing committed.

v2 — reviews applied (2 October 2026)

Skill: course-build → write-lesson (article lines) → write-video-script (L02's beats) → write-mcqs (every item change; check_key_tells.py) → write-explanation (fb / why lines) → sentence-atomise (every new sentence) → review-figures (author re-render of every changed figure) @ efd8fba

In plain words. Three reviews were applied in one pass, in their order of authority: the blind read (every key agreed; its flags table), the guidance / questions / accuracy review (two Essentials and every Worth-fixing row), and the figure review (every chip now sits on its thing or is leadered to it). L02 no longer re-teaches Grade 3's reservoir lesson: it recalls the word "reservoir" in one line and its rule word for word, the lawn-and-hose scene is gone (the town has turned off its fountains), and the article is 294 words. The bank's keys no longer give themselves away: KEY TELLS: 0 of 122 MCQs flagged (a: 0, b: 0, c: 0). The model drawing's cup is open, the falling drop sits above its rim, and every LIQUID / GAS / SOLID chip is on the water, the dots or the ice it names. Nothing is committed and no video batch was sent.

#### Finding → change

#### Questions for James — built as recommended

1. 97 / 2 / 1 kept (USGS 96.5 / 1.74 / 0.76 rounds to it); "water under the ground" in plain words. 2. "about 9 of 10" kept. 3. Both desalination sentences kept (one line now). 4. The cold spot kept, with the one solid-water line (row 14). 5. L05 kept whole (413 words; a figure or subtitle every ≤ 6 lines). 6. No action on Topic 23's chips.

#### Key lengths

python3 ~/Documents/CourseBuilding-Vault/skills/write-mcqs/check_key_tells.py grade4/topic24_build/content/G4T24.py → KEY TELLS: 0 of 122 MCQs flagged (a: 0, b: 0, c: 0) (KEY LONGEST any margin: 24 of 122; fixed sets exempt: 38). Own script (key length against the longest distractor, every MCQ): bank: 50 MCQs, 0 keys the single longest; lesson / practice / summary: 71 MCQs, 23 keys longest by any margin (none beyond the gate's tolerance; the Essential was scoped to the bank). KEY_TELL_EXEMPT_SETS not needed: PLACE4 / SOURCE4 / STEP4 are exempt by appearing under two or more items.

#### Counts and key strings (v2)

• Bank 57: 22 easy · 18 medium · 17 hard; per atom 058 14 · 059 9 · 060 10 · 061 11 · 062 13; sorted easy → hard, atoms rising inside each tier (asserted); 50 MCQ · 6 typed · 1 multi; MCQ keys A 10 · B 18 · C 12 · D 10.
• Quizzes: L01 (11) A A B B B C A B B A C · L02 (9) B A C B B A C B B · L03 (9) A A C B A B A B C · L04 (9) A A B C B B C A C · L05 (12) B A B B C A C B A A A B. Practice (15, 13 MCQ) A B C A A C B B A C A C B. Attached, video order: A A B B A B C C.

#### Checker results

• build_tools_check_topic.py → 0 FAIL, 33 WARN (S8 beats of 3–5 sentences, kept; S36 18 words, all answered in v1 and unchanged — "reservoir" and "fountains" are not flagged). S38 fragments 0.
• grade4/check_g4_articles.py → 0 FAIL: articles L01 295 · L02 294 · L03 298 · L04 320 · L05 413 (WARN above 300 kept, as before); scripts 150 / 149 / 150 / 148 / 148; intro 118; summary 175, opener exact.
• build_tools_names.py Ortiz Leah Levi Liana → all free, exit 0.
• check_svg_ids.py figures.py → 43 figures, ids 0, PASS.
• build_tools_build_topic_site.py … /tmp/g4t24_v2.html → 5 lessons, 15 practice, 57 pp100, 43 figures; each lesson's article route screenshotted at 900 px and read (L01–L05) plus the L05 quiz route; every figure renders in place.
• build_tools_video_batch.py + check_brief.py → 6 of 7 clean (the intro's standing hit) — not sent.
• Figures re-rendered to /tmp/g4t24_figs/ and read on four contact sheets: every chip on its thing or leadered to it; the cup open with the drop above its rim; the tub distinct from the bowl; the grid leaders clear of the green square; the lake chip under the lake; the salt chip inside the ocean; the rule cards word for word.
• Blind bundle blind/2026-10-02-v2/ (121 MCQ, 7 typed, 1 FRQ, 1 multi; keys sequestered). grep "\[Drawing:" | grep -ci "salt\|fresh\|most\|ocean\|frozen\|drink\|solid\|liquid\|gas" → 0 hits.

#### Items changed (stem, options, key, model, figure, blind text or accept list)

• Lessons: L01-04 (fresh-water term moved up, "sea water" distractor), L01-05 (distractors lengthened), L01-06 (ice-cap term moved up), L01-08 (claim flipped: ice caps are water — yes), L01 old q9 cut, L01-11 (grid now last; figure q_grid_jkl leaders), L01 article (+globe_photo); L02 headache / article / script / forest / bullets / rule, L02-V (fb, why), L02-01 (key wording), L02-03 (reservoir), L02-04 (stem, options, fb), L02-05 / 06 / 07 (fb, why; q7 key), L02-09 (options); L03-01 (new recall), L03-09 (the why item, moved), L03 check_design; L04-04 (key, fb), L04-06 (stem, fb, why), L04 rule card; L05-01 (new recall), L05-05 (stem), L05-09 (figure q_model_jklm), L05-10 (stem; figure q_model_before_shade), L05-11 (figure q_model_before_noice), L05 article (+solid-water line, captions, "Labeling"), L05 rule card.
• Practice: PR-07 (why), PR-11 (figure pr_model_stuv, blind text), PR-14 (options, fb).
• Summary: beat 3 and beat 5 (lexicon wording), bullet 2.
• Bank (v2 numbering): every item renumbered; changed — PP-1, 5, 6, 7, 8 (figure pp_grid_nrmp, stem, letters, blind), 9, 10, 11, 12 (new), 13 (retagged, options), 16, 17, 18, 19, 20, 22, 23, 24, 25, 26, 27, 28, 29, 33, 34, 35, 37, 38, 39, 40, 41 (new, typed), 43, 45 (stem, figure pp_ocean_rain_nprq chips and blind), 46, 47, 48, 49 (figure pp_tub_qprn, letters, blind), 53, 54, 55. Unchanged: PP-2, 3, 4, 14, 15, 21, 30, 31, 32, 36, 42, 44, 50, 51, 52, 56, 57.
• Figures deleted: grid_plain, model_plain, model_start, ocean_rain_plain, table_dry_summer, hose_cartoon. Added: fountain_cartoon, pp_grid_nrmp, pp_tub_qprn (pp_grid_npq and pp_model_rnpq retired). Changed: model_parts, model_states, model_steps, model_after, q_model_before_shade, q_model_before_noice, q_model_jklm, pr_model_stuv, build_steps (the cup), q_grid_jkl, fresh_zoom, ocean_rain_labels, pp_ocean_rain_nprq, salt_dish_strip, rule_matters, rule_connects, rule_model; beach_cartoon and build_cartoon briefs.

#### Sentence-atomise, the new lines

L02's five reservoir sentences: one idea each, the actor named (the town, the rain), the cause before the effect ("using … faster than the rain replaces it, so the reservoir is running short"), the plain phrase before the term ("a reservoir: a big lake built to store a town's water"). The solid-water line in L05 names the thing ("the snow and the ice of the ice caps and glaciers"). Every new fb line states the option's error, then the rule. S1 / S2 / S38 clean.

v3 — polish (2 October 2026)

Skill: write-mcqs (Reader C's six notes) → write-video-script (one L01 beat) → sentence-atomise (every new sentence) @ fb37d14

In plain words. Reader C read the whole v2 bundle cold and agreed every key (130 items). Two items needed a fix and four needed a check; all six are done. The L01 video now says where the one reachable square sits ("nearly all under the ground"), so PP-04's key is taught in the video as well as the article; the grid item no longer sits under the item whose stem and feedback state its answer; the mountain-lake item asks about a cup of water, not a drop; the dew item is a new move (PP-31's mirror); the Greenland item no longer repeats the practice set's ice-cap stem. Keys keep their letters. Nothing is committed and no batch was sent; the L01 beat changed, so the video batch is rebuilt before it goes.

Changed ids. L01 script (beats 1 to 3 and beat 3's screen cue) · L01 check_design · L01-07 and L01-08 swap places (the grid item is now L01-07; the bundle's ids L01-07 to L01-11 follow the new order) · L04-06 (stem, fb, why) · PP-28 (stem, options, why; key letter D kept) · PP-41 (stem). Type sheet: the Type 13 exemplar, inventory row 28, the ⟂ list. Nothing else touched.

Gates.

• python3 build_tools_check_topic.py grade4/topic24_build/content/G4T24.py → 0 FAIL, 32 WARN (the same S8 beat-length and S36 word WARNs as v2, all answered in v1; S30 typed=7 fail=0 warn=0; S38 fragments=0; S39 hits=0).
• python3 grade4/check_g4_articles.py grade4/topic24_build/content/G4T24.py → 0 FAIL: scripts 150 / 149 / 150 / 148 / 148; articles 295 / 294 / 298 / 320 / 413 (the two over-300 WARNs as before); intro 118; summary 175, opener exact.
• python3 build_tools_build_topic_site.py grade4/topic24_build/content/G4T24.py /tmp/g4t24_p.html → 5 lessons, 15 practice, 57 pp100, 43 figures.
• python3 ~/Documents/CourseBuilding-Vault/skills/review-figures/check_svg_ids.py grade4/topic24_build/content/figures.py → figures read 43, ids found 0, PASS.
• python3 ~/Documents/CourseBuilding-Vault/skills/write-mcqs/check_key_tells.py grade4/topic24_build/content/G4T24.py → KEY TELLS: 0 of 122 MCQs flagged (a: 0, b: 0, c: 0).
• python3 build_tools_names.py Ortiz Leah Levi Liana → all free, exit 0.

Bundle. blind/2026-10-02-v3/ (121 MCQ, 7 typed, 1 FRQ, 1 multi; 130 held; keys sequestered). grep "\[Drawing:" | grep -ci "salt\|fresh\|most\|ocean\|frozen\|drink\|solid\|liquid\|gas" → 0.

Owed after this. The video batch rebuilt from the changed L01 beat (build_tools_video_batch.py, then check_brief.py) before it is sent on James's word; the picture pass after his answers; the Needs-you page records/REVIEW_NEXT_G4T24.md → course_preview/review/REVIEW_NEXT_G4T24.html (written in this pass).

Skill: course-build → review-lesson (Roles B, B2) → review-questions → review-accuracy (non-author pass; filed by the coordinator from the reviewer's hand-back) @ efd8fba

Topic 24 — guidance, questions and accuracy review, 2 October 2026

The science is honest at Grade 4 and every claim re-ran true except four: L02's middle section re-teaches Grade 3 Topic 28's lesson almost scene for scene (dry August, a hose on the lawn, the town's reservoir filled by spring rain, pipes to every house) and drops Grade 3's coined word "reservoir" for "a lake"; the bank's keys are test-wise bait (26 of 49 MCQ keys are the single longest option; in 8 items the key is the only option with a second clause); the recall of Topic 20's glacier line is a recap form, not word for word; L02 is 339 words, not 337. Lettered stems match their drawings in all six figures; blind lines carry data only; counts true; names clean; Topic 23's recalled sentences match its lexicon and lesson text word for word.

The author's six questions, the reviewer's view: (1) keep 97 / 2 / 1 (USGS 96.5 / 1.74 / 0.76 rounds to it honestly; "water under the ground" in plain words, Grade 5 coins the term); (2) keep "about 9 of 10"; (3) keep both desalination sentences; (4) the cold spot is right — add one line that in the real cycle the solid water is the snow and the ice caps; (5) keep L05 whole; (6) no action on Topic 23's chips.

Skill: course-build → review-figures (non-author pass; filed by the coordinator from the reviewer's hand-back) @ efd8fba — figures at g3-5-science commit 4761525

Topic 24 — figure review, 2 October 2026

The hundred-squares grid is excellent and counts true (97 blue, 2 white, 1 green on the render). The model is clear and sturdy, but its chips float: in the states, steps and lettered forms the labels for the drops, the ice and the cup's water sit on empty sky, among the vapor dots, or under the table, so a child reading LIQUID beside dots and GAS beside dots sees two identical regions with opposite names. Page pass: every figure renders on the built page as alone; check_svg_ids.py PASS; Topic 23's imported drawers byte-identical to Topic 23's (its three top chips tight but not overlapping).

1. Model chips off their things (model_states, model_steps, q_model_jklm, pr_model_stuv, pp_model_rnpq): the drops chip sits among the vapor dots; the ice chip floats above the rim; the cup chips sit under the table. In q_model_jklm J and K both read as marking dots (a second defensible option); in pp_model_rnpq N is among dots; pr_model_stuv's alt says "U on the bag of ice" and it is not. Fix in model_body: each chip on or leadered to its thing; verify by render.
2. The falling drop sits on a closed cup (every model figure). Draw the cup open at the top; the drop above the rim, under the sag.
3. ocean_to_rain: "A MOUNTAIN LAKE" and chip Q are in the soil, 60 px below the lake; blind text and alt say "beside the oval" — false. Move to the lake. Same figure: no Sun → water energy arrow while a grey vapor arrow rises straight at the Sun — the "Sun pulling" reading the language rule forbids. Import Topic 23's yellow arrow to the ocean; shorten or drop the grey arrow.
4. fresh_zoom clips text: "a tiny slice of the square" is below the frame and absent on the page; the title is edge-clipped. H → ~235, title right.
5. A caption says "real" (L05, model_drops_photo: "A real model after an hour in the sun"). "A classroom model…".

Unplaced (5): globe_photo — PLACE (intro beat 2 or L01); grid_plain, model_plain, model_start, ocean_rain_plain — DELETE.

Minor: K / N leaders cross the green square's corner; pp_grid_npq's R line short of the grid width; salt_dish_strip water corners outside the dish; model_parts CUP chip hides the drop, COVER floats; collected water drawn green (say once that green means fresh); "THE SALT STAYS IN THE OCEAN" chip straddles the ocean's edge; left/right on beach_cartoon and build_cartoon briefs.

Three best: grid_labels, ten_cups, build_steps. Three weakest: model_states, pp_ocean_rain_nprq, fresh_zoom.

Skill: course-build → review-lesson (Role A blind key read, two readers; Role C adjudication by the coordinator) @ efd8fba

Topic 24 — Where Earth's water is — blind read, 2 October 2026

Every key agreed by two blind readers (127 items each; MCQ disagreements 0 for both; PP-54 multi A, D matched; the six typed words on the lists; the FRQ model in the key's words). The key both readers worried about, PP-04 "Where is most of the fresh water that is a liquid?", IS "under the ground" — the readers answered it so and the key agrees; the topic teaches ground storage (PR-06, PP-24). All lettered stems match their drawings; no drawing description gives its answer away. The intro poll is not bundled by design.

Flags carried into the fix pass

Readers' files: blind/2026-10-02/READER_A.md, READER_B.md.

#### Reader C, the v2 re-read (blind/2026-10-02-v2/READER_C.md)

Skill: review-lesson @ f9c548d (Role A — blind key reader, Reader C, cold read of BLIND_BUNDLE_G4T24.md v2 only)

Takeaway

The bundle answers cleanly. I could answer every item from the bundle alone and found no item with two defensible options and none with zero. The keys are not spottable by length or shape: the yes/no verdicts run mixed (no, no, yes, no, yes, no, yes, no, no, yes, no, no, yes), and the few longer keys (L04-02, PP-10) are longer because they are specific, not because they carry a "because". Three things to check against the lessons I cannot see: PP-04 keys "water under the ground" as where most LIQUID fresh water is, but the L01 grid (97 / 2 / 1) has no groundwater chip, so a student taught only the grid picks "rivers and lakes" — the lesson must say it; L01-11's answer is handed over by L01-07 on the same page (about 1 square = fresh liquid, the one green square); and "glacier" appears in L01-05/06 before anything in the bundle defines it. PP-20 (fog) and PP-28 (dew) key "the ocean" by the course's 9-in-10 rule — right for the model, a little loose for the science. Only one child is named (Levi); the drawings' descriptions name counts and colours but the drawing itself shows the same counts, so no description hands over more than the picture would.

Items

Adversarial answers (typed items)

Cross-item notes

• Same-page giveaway: L01-07 (about 1 square = fresh liquid) answers L01-11 (the 1 green square). Move L01-11 before L01-07, or ask L01-11 for the frozen chip instead.
• Groundwater: PP-04 keys it, PP-13 and PR-06 lean on it, but nothing in the bundle's L01 items or its grid names it. If L01 does not teach "water under the ground", PP-04 fails; PP-13 survives (the sea and sand options are absurd).
• Terms before coining: "glacier" (L01-05, L01-06) — not defined anywhere in the bundle; "reservoir" is glossed in its stem; "water vapor" and "condensation" are defined in their own first stems; "dissolved" is shown by L04-01 before L04-09 uses it.
• Repeated stems: PR-12 = L01-04 and PR-13 = L04-09 with four options (fine for practice recall); PP-41 is PR-15 with the pole swapped; PP-20 and PP-28 are one design move (fog/dew) twice.
• Shape cues are faint: L04-02 (only comma), PP-10 (longest), PP-46 (key first and canonical). No key is the only option with "because/so/and"; capitals and full stops are consistent within each item.
• Verdict string on "Is she/he right?" items: no, no, yes, no, yes, no, yes, no, no, yes, no, no, yes — mixed.
• Names: Levi only; no two-name confusion. No scene reads as pasted from another topic; all are water scenes.
• Wording: L04-06 "a drop of water floats in a lake" is the one un-exam-like stem.

Would not ship as is: PP-04 (unless L01 teaches groundwater — coordinator to confirm against the lesson); L01-11 (same-page giveaway by L01-07 — reorder or retarget). Everything else ships; L04-06 wording and the "glacier" coining are check-and-fix items, not blockers.

Skill: course-build → write-mcqs → write-explanation @ 3a2c39d; v2 @ efd8fba

Topic 24 — question types in the PP100 bank, one exemplar each (2 October 2026, v2 after the reviews)

James reviews a bank as its question TYPES first, one exemplar each (vault: 50 Decisions/A bank is reviewed as its question types first, one exemplar each.md). This sheet lists the eleven types the Grade 4 Topic 24 bank runs on and prints one exemplar per type in full as it ships in content/G4T24_pp100_bank.py. v2 (2 Oct 2026, evening): the blind read, the question review and the figure review applied — every key one clause and never the single longest option (check_key_tells.py: KEY TELLS 0 of 122), every stem a positive question, US words, named farms, the tub model's moves and readings varied, the bank's two drawn items on their own drawings (a grid filled from the other corner; the TUB, not the bowl), two EAR4-058 term recalls added, the cactus-farm item retagged to EAR4-058.

One cluster. The topic is one PP100 cluster of five atoms (EAR4-058 → 059 → 060 → 061 → 062, capstone 062). Items per atom: 14 · 9 · 10 · 11 · 13 = 57. Tiers: 22 easy · 18 medium · 17 hard (6 typed, 1 multi-select, 2 drawn items, 8 four-option MCQs with a "hard because" line each). EAR4-058 carries easy items and its two term recalls (one typed, hard tier); EAR4-059 easy and medium; EAR4-060 and 061 run from easy to hard; EAR4-062, the capstone, medium and hard only. MCQ keys A 10 · B 18 · C 12 · D 10 (50 MCQs).

The bar list (things a generated bank must never use because they break a taught claim or a rule): "gravity" as the puller ("Earth pulls the drops down" is Grade 3's sentence); "the Sun lifts / pulls / sucks the water up" (nothing is lifted: the Sun's light carries energy, the water warms up and evaporates — EAR4-056's cross note); "water is a form of energy" or "the Sun's energy is in the water"; "groundwater", "aquifer", "desalination", "percent" as student words (this topic says "water under the ground", "take the salt out", "of every 100 squares"); "iceberg" or any word no lesson taught as an option; the Great Salt Lake or any salt lake as an instance of "a lake" (the topic's lakes are fresh); a sea fish, seal or sea bird as a living thing that "needs fresh water" (the rule says "on land"; the sea fish is the table's exception row); tasting as an observation (the cup's water is STATED to be fresh, never tasted); "bowl / cup / plastic wrap / bag of ice" as a bank model (the lesson's model; the bank's model is a tub, a jar, a glass lid and a tray of ice cubes, drawn as such); any lesson, quiz, attached or practice scene as a bank scene; "Farm A / Farm B" or any lettered instance; "taps", "washing-up" or any British word; a stem of the form "Which sentence shows he is wrong?" (every stem is a positive question); a key that is the single longest option or the only option with a second clause; "strand"; "biggest / smallest" for a count or a share.

⟂ draw constraints (pairs never served in one PP100 draw: one item states what another asks):

• PP-2 (97 squares are in one place: the ocean) ⟂ PP-1 (what is nearly all of the blue) ⟂ PP-9 (the true sentence about the blue).
• PP-3 (where most fresh water is) ⟂ PP-8 (the frozen water's letter on the grid) ⟂ PP-11 (the true sentence: most fresh water is frozen) — near-duplicates, draw one.
• PP-21 (9 of 10 cups come from one place) ⟂ PP-30 (how many of the 10 cups come from everywhere else) ⟂ PP-42 (type the place most water evaporates from).
• PP-20 (fog's water) ⟂ PP-29 (the true sentence about rain 500 miles from the sea) — one source item per draw; PP-26's stem states the source, so PP-26 is never drawn with either. PP-28 (every lake and river stops evaporating: only a little less rain) ⟂ PP-31 (the ocean stops evaporating: much less rain) — the two stopped-evaporating items, draw one (v3).
• PP-22 (what is left in the salt pans) ⟂ PP-44 (type the white stuff left on the rock) ⟂ PP-36 (which step leaves the salt behind).
• PP-38 (the tub's drops stand for a cloud) ⟂ PP-49 (which letter on the drawn tub stands for a cloud).
• PP-39 (the ice cubes are the solid) ⟂ PP-56 (select every liquid part) — the second's distractors name the first's key.
• PP-37 (the tub's salt water is the ocean) ⟂ PP-52 (type the real thing the warm salt water stands for).
• PP-50 (type the state of the vapor: gas) ⟂ PP-40 (the state of the water in the air under the lid).
• PP-12 (the harbor's water is salt water) ⟂ PP-5 (which of these is salt water); PP-41 (type ice cap) ⟂ PP-7 (the ice of a glacier is the frozen fresh water).

The eleven types

(Type numbers 5 and 8 were merged into 4 and 9 during drafting; type 13 was added in v2 for the term recalls.)

The exemplars, in full

Type 1 — PP-2 (easy, EAR4-058). Of every 100 squares of Earth's water, 97 squares are in one place. Which place?

Options: the ocean · ice caps and glaciers · water under the ground · the rivers and the lakes.

Why: Of every 100 squares of Earth's water, 97 are salt water in the ocean. So the place is the ocean.

Type 2 — PP-5 (easy, EAR4-058). Which of these is salt water?

Options: the water in a clear forest river · the ice of a glacier · the water a sea turtle swims in · rain falling on a town.

Why: Salt water is water with salt dissolved in it, and the ocean's water is salt water. A sea turtle swims in the sea. So the water a sea turtle swims in is salt water.

Type 3 — PP-8 (easy, EAR4-058, figure pp_grid_nrmp). The grid shows all of Earth's water as 100 squares. Which letter marks the fresh water that is frozen?

Options: N · R · M · P.

Why: Fresh water that is frozen is 2 of every 100 squares: the ice of the ice caps and glaciers. R marks the 2 white squares. So R marks the frozen fresh water.

Type 4 — PP-9 (easy, EAR4-058). A boy says Earth looks mostly blue from space, so there is plenty of water for everyone to drink. Which sentence is true about the blue water?

Options: It is mostly ice. · It is mostly fresh water in lakes. · It is nearly all salt water. · It is mostly water vapor..

Why: Nearly all of Earth's water is salt water in the ocean. Drinking salt water makes people sick. So the blue is not water people can drink, and the boy is not right.

Type 6 — PP-17 (easy, EAR4-059). Why does the small share of fresh water on Earth matter so much to people?

Options: People and farms can use only fresh water. · Fresh water is heavier than salt water. · Fresh water is the only water that evaporates. · Fresh water is the only water in the ocean..

Why: People, animals and plants on land need fresh water. Only 1 square of every 100 is fresh water we can reach. So the small share of fresh water matters to every person and every farm.

Type 7 — PP-26 (medium, EAR4-060). A heavy rainstorm soaks a prairie 1,000 miles from the sea, yet most of that rain's water evaporated from the ocean. How did the water get so far from the sea?

Options: Rivers carried the water uphill to the prairie. · The wind carried the clouds across the land. · The rain soaked up through the ground from the sea. · The ocean rose over the prairie..

Why: Most of the water in the air evaporated from the ocean. The wind carries water vapor and clouds a long way across the land. So the ocean's water reached the prairie as rain.

Type 9 — PP-33 (medium, EAR4-061). A drop of water in a farm pond fell as rain last night. Where was most of that water before it went into the air?

Options: in the pond all along · under the farm · frozen in a glacier · in the ocean.

Why: Most of the water in the air evaporated from the ocean. Evaporation left the salt behind, and the water fell as rain into the pond. So most of the drop's water was in the ocean.

Type 10 — PP-38 (medium, EAR4-062). After a morning in the sun, the class lifts the lid of the tub model and finds drops hanging under it. Which real thing do those drops stand for?

Options: the ocean · a cloud · a river · a glacier.

Why: A cloud is tiny drops of water that condensed high up where the air is cold. The drops under the lid condensed under the cold spot. So the drops stand for a cloud.

Type 11 — PP-53 (hard, EAR4-062). A tub model is filled with cold salt water instead of warm, and it stands in the shade all morning. What is in the small jar at noon?

Options: a jar full of fresh water · almost nothing · a jar full of salt water · a lump of ice.

Why: The Sun's light warms the water, and warm water evaporates. Cold water in the shade hardly evaporates at all, so almost no vapor condenses and almost nothing drips into the jar. So the jar holds almost nothing.

Type 12 — PP-42 (hard, EAR4-060, typed). Most of the water that goes into Earth's air evaporates from one place. Type the name of that place.

Accept: ocean · the ocean · oceans · the oceans · sea · the sea · the seas · ocean water · sea water.

Why: Most of the water that goes into the air evaporates from the ocean. That is because the ocean is where most of the water is. So the place is the ocean.

Type 13 — PP-41 (hard, EAR4-058, typed). A plane flies over Greenland, far in the north near the North Pole. Below it, ice more than a mile thick covers nearly all the land. Type the two-word name for this ice. (v3: re-scened; v2's stem was PR-15's with the pole swapped.)

Accept: ice cap · an ice cap · the ice cap · ice caps · icecap · ice-cap · the ice caps.

Why: A thick sheet of ice that covers the land near the North Pole or the South Pole has a name. That name is an ice cap.

Design-move inventory (every bank item: the move, the scene kind, the nearest lesson-side instance)

Lettered renders: pp_grid_nrmp (N, R, M, P; the grid filled from the top-left corner, chips above the three squares, the whole-grid line along the bottom) — PP-8 only; pp_ocean_rain_nprq (N on the ocean, P on the dots, R on the rain, Q on the lake) — PP-45 only; pp_tub_qprn (the TUB: Q on the dots, P leadered to a drop, R above the tray of four cubes, N on the water) — PP-49 only; the quizzes use q_grid_jkl (J, K, L) and q_model_jklm (J, K, L, M); the practice uses pr_model_stuv (S, T, U, V). No lettered render is shared between the quizzes, the practice set and the bank; no letters A–D. The counts on every grid stay 97 / 2 / 1: they are the data.