# Desk run-through · Foundation mathematics Weeks 37–38

**29 September 2026 · authored draft, no teacher, child or class observed.** I used the [scripts](LESSONS.md), [child cards](LEARNER-CARDS.md), exact [source recipes](MATERIALS.md), [fresh checks](STUDENT-CHECKS.md), separate [public key](teacher/KEY-AND-NEXT.md) and all eight print pages in an offline **desk simulation**. I did not physically print or cut cards, observe a child, trial AAC/tactile access, validate 25-minute classroom pacing or measure learning. The examples below are checks of artifacts and plausible actions, not outcomes.

| Role and case | What I actually did | Finding, repair and limit |
| --- | --- | --- |
| Teacher, Day 183 | Read the two **given** questions separately; enumerated the ten source card labels from the print sheet. First sort gives BUS 2+4=6 versus BIKE 1+3=4. Reset exactly the same ten; second sort gives STAR 2+1=3 versus PLAIN 4+3=7. Both totals equal ten. | The script explicitly resets rather than treating the two tallies as 20 distinct observations. The child is not asked about their own transport. A real teacher must check paper handling and the group's attention span. |
| Teacher, Days 186–189 | Traced P186 as fourteen objects against a false 15 card, P187 as 7 spread versus 11 compact, P188 as 5+3=8 against false 7 and P189 as 8−2=6 against false 5. In the subtraction model, six remain IN and two stay OUT, accounting for eight. | The object model, not visual row length or printed claim, controls the result. Recheck language permits “confirmed” if a child's first response is already correct. No actual student explanation has been seen. |
| Learner route, Day 182 | Followed A (pair), B (one mark per source) and C (direct adult) using the same invented 5 SUN/7 CLOUD source. Each route ends with two unmatched CLOUD pictures or an equivalent 7 versus 5 count. | All routes demand the same sort, two counts, comparison and reason; none depends on speech, writing, cutting, hearing or colour. A particular child's access requires local trial. |
| Learner, fresh Check A **before** reading the key | Used the child prompt and Check A sheet only. Moved each of thirteen printed picture sources once into ARC/BAR rows and kept a mark trail. Then read the separate key and compared its 9/4, five unmatched ARC result to the physical icons. | The child sheet and prompt do not state the aggregate answer or link the key. The result is limited to the supplied fictional sheet. The key and source remain public by URL, so earlier access can affect freshness. |
| Learner, fresh Check B **before** reading the key | Counted the compact A row and spread B row on the rendered page, paired eight marks, then modelled the separate START and ADD frames before looking at the display claim. Compared to the key: 12/8 leaves four A unmatched; 4+5 is nine, so displayed eight needs repair. | The two source stories are distinct from all routine numeric cases and swaps. A right first prediction is recorded as confirmed, not as a fictional mind change. A physical spacing/tactile layout may differ after local production. |
| Time-poor supporter | Tried the no-cost drawn route: four broad ring marks versus six wide line marks for an invented page, then a separate START of three paper tickets with two added. No internet, purchase, personal story or real-world survey is necessary. | The adult can present a given question and let the child direct marks. This is a desk script, not evidence that home engagement occurred. |
| Potential funder / rights reviewer | Followed four exact Foundation maths codes and row numbers from the dated [ACARA snapshot](source-snapshot.json) to verbatim [crosswalk](CURRICULUM-CROSSWALK.md), then separated ACARA words from original scenario/data/print rights in the [ledger](SOURCE-AND-RIGHTS.md). Checked the public formative-key and partial-coverage wording. | This is a reviewable ten-day sample within a planned year sequence. It does not establish a complete year, state approval, funding eligibility, classroom effect, translation review, First Nations content, accessibility certification or a secure assessment system. |

## Print and automated evidence

- Generated **eight original one-page physical A4 SVG/PDF aids** using [local vector code](print/generate_print.py). Rendered every PDF to PNG and visually inspected an eight-page contact sheet. The first review caught word-only labels on the route and Check A cards; I added distinct original bus/bike, star/plain and arc/bar pictograms, rerendered and inspected those two sheets again. Titles, dark icon/card outlines, group labels, footer and all check objects stayed within the page. PDFs have selectable text, but are **untagged**; [full text/tactile recipes](print/TEXT-ALTERNATIVES.md) are the adaptation source. No physical print or assistive-technology test has been performed.
- Counted SVG source attributes, not colour: Day 183 two-way ten cards; Check A 9 ARC/4 BAR; Check B 12 compact A marks/8 spread B marks and 4 START/5 ADD icon cards; twelve broad blanks. The print sheet does not display a Check A category total or Check B corrected result. The local [verifier](verify_pack.py) repeats these data checks and checks `pdfinfo` and `pdftotext`.
- Ran Ruff over **all new-pack Python** with `E,F,I,UP,EXE,PLR,RUF`, Ruff format check and Python compile check; all pass. The [verifier](verify_pack.py) checks official workbook SHA, exact row/level/wording, ten six-phase 25-minute scripts, thirty same-target routes, twenty worked swaps, both fresh checks, source arithmetic, print structure, local links and the [SHA-256 manifest](manifest.json). Its result is an internal consistency receipt only.

## Educator follow-through before use

Check the applicable Australian state/territory and school sequence; print a page at actual size; keep objects broad and safe; try neutral read-aloud/AAC/tactile/directing routes with real learners; allow a child to confirm an accurate first idea; collect authentic first responses privately; and vary the next case based on the child's own sort/model trail. Do not publish real child work or infer population facts from these invented cards.