How did people keep food cool before refrigerators?
Labs reference: How did people keep food cool before refrigerators?.
Lesson: A useful answer needs a place, a time, a purpose, and the people who had access.
Provenance: AI-assisted RS Learning Lab teaching example, researched and drafted in Codex with web search on 2026-09-10. The prompts and first draft below are constructed teaching material, not a transcript of a learner session. The comparison board is a source-based synthesis. No classroom observation or historical experiment is claimed. Human editorial and pilot review remain open.
Return to the Gallery · Open the Question Path
1. Start with the question
Open with this, before showing the comparison board:
How did people keep food cool before refrigerators?
Let people offer a first thought: ice, a cellar, a stream, a clay pot. Then ask: Would your answer work in an ancient desert settlement? A palace? A rural household? A crowded city?
Keep the inquiry about historical cooling and access. We are comparing evidence, not testing food-storage methods or deciding whether something is safe to eat.
2. Give AI one useful job
Use a conversational AI tool with web search and source links. Its job is to help divide the question into researchable cases and find evidence. Open the sources yourself. With a conversation-only tool, provide selected source excerpts and their links, and ask it to compare those; do not ask it to pretend it has searched.
First prompt to try live:
Help me investigate how people kept food cool before mechanical refrigeration. Start with a few possible methods and the questions I should ask about them. Keep this a historical inquiry. Do not assume one method worked everywhere or was available to everyone.
For a short opening demonstration, a facilitator can operate the tool while learners choose the next question. A prepared page alone does not replace the lab’s live-AI encounter.
3. Notice what the first answer leaves out
Constructed first draft for discussion:
People used ice houses, spring water, and clay pots to keep food cool before refrigerators.
This gives us search directions, but no place, date, source, or account of access. It also treats storing ice, chilling a drink, and keeping food cool as the same job.
Follow-up prompt:
Split this into named places and periods, including ancient times, desert communities, palaces, rural households, and cities. For each case, identify what was cooled, the method, and who could use it. Link evidence I can inspect. Separate contemporary records, surviving objects or buildings, later historical interpretation, and modern experiments. Mark gaps rather than filling them with a plausible story.
Add a question about labor: Who collected the ice, maintained the water supply, carried deliveries, or cared for the food? If the source does not answer, keep that gap visible.
4. Build a comparison board from sources
Read this after the group’s first attempt. These are selected cases, not a complete global history. Later examples do not establish what happened in ancient times. The modern case at the end tests our assumptions about the timeline.
Scroll sideways to see all columns.
| Place and period | What the source supports | What to ask next |
|---|---|---|
| Mari, middle Euphrates, 18th century BCE — ancient court | Jørgensen reports texts describing mountain ice brought to court, apparently chiefly for royal wine. Iranica, “Literary evidence” | What evidence would reveal ordinary households’ practices? Court records do not answer that. |
| Iranian plateau — desert-edge communities, documented later ice houses | Many yaḵčāl systems froze water in winter basins and stored the ice for summer distribution. Operators included cooperatives and private owners. This was ice storage, not a food-storage room. Iranica, “Use and typology” | How did season, water supply, labor, and ownership shape access? Do not assign all surviving domes an ancient date. |
| Versailles, France — royal estate, Louis XIV and the 18th century | Insulated pits held winter ice for summer drinks, sorbets, and ice cream. Palace of Versailles, “the ice stores” | Who supplied and maintained this service? The palace example does not establish ordinary households’ access. |
| Hopewell Furnace, Pennsylvania — industrial community, springhouse rebuilt around 1806–1807 | Food containers stood partly in flowing spring water. The cooling room was locked and reserved for the ironmaster’s mansion. NPS, “Spring House” | A resource existed nearby, but who could use it? What did other households do? |
| Walker family homestead, Great Smoky Mountains — late 19th and 20th centuries | A creek-side springhouse cooled milk and butter; the farm also used a smokehouse and stored pickled vegetables. NPS, “The Walker Sisters,” “Early Days on the Homestead” | Cooling, curing, and pickling are different practices. Which local resources and household work supported each? |
| American towns and cities — late 19th and early 20th centuries | Household iceboxes used delivered blocks of ice. The Smithsonian documents harvesting, transport, insulated boxes, and delivery cards. National Museum of American History | An icebox depended on a supply service. What did price, delivery coverage, housing, and interruptions mean for a particular neighborhood? |
| Mopti region, Mali — modern comparison, research and training reported for 2016–2021 | MIT D-Lab, the World Vegetable Center, and Institut d’Economie Rurale studied clay-pot cooling and worked with farmers, sellers, and pot makers. The devices provide a cool, humid environment for vegetables in dry regions. MIT D-Lab, “Clay Pot Coolers” | Why might a method without mechanical refrigeration remain useful today? Modern performance evidence is not evidence of an ancient invention date. |
In the last column, questions are ours; they are not claims that the linked sources have answered them. Look for the people maintaining the system as well as the person credited with owning it.
5. Check a claim that sounds satisfying
Try this claim as a constructed checking exercise, not a quotation from an observed AI response:
Ancient Persian ice houses were refrigerators that kept everyone’s food cold.
Open Jørgensen’s Iranica article, pages 1–3. It distinguishes ancient storage-pit evidence from the uncertain development of later ice-house buildings. It does not justify a universal date or universal household access. Revise the claim to match the evidence, rather than replacing it with a different sweeping story.
For a second check, open Hopewell’s source and find who could enter the cooling room. That detail changes an answer about technology into a question about access.
Prompt after reading:
Using only the passages we have checked, help revise our comparison. Separate what the sources support from our interpretations and unanswered questions. Do not generalize from a palace or one household to everyone in that society. Show what you changed and why.
The AI can help organize this revision. You decide whether its wording stays within the evidence.
When sources give different descriptions
Our source check found a disagreement in the description of Mari. Iranica’s “Literary evidence” section, on PDF page 3, places the Kingdom of Mari on the “upper Euphrates.” The official excavation history places Tell Hariri, the city site, in the “middle valley of the Syrian Euphrates.” These are statements in the sources, not invented competing answers.
Open both passages. Identify what each describes: the kingdom or the excavated city. Then consult the excavation site’s geographical location page. What evidence would help us decide whether the different wording reflects geographic conventions, scope, or an error?
For the comparison board, we use middle Euphrates, following the site-specific excavation account. We have not established why the descriptions differ. Naming that uncertainty does not make the two descriptions equally well supported for locating the city, or invalidate everything else in either source.
Prompt after comparing:
These sources use different geographic descriptions. Separate what each says from possible explanations for the difference. Help me write a sentence locating the city, name the source supporting it, and state what I have not resolved. Do not invent a reconciliation.
6. Make a better answer, then choose where to go
Final teaching synthesis:
There was no single solution or worldwide changeover. Our examples use stored winter ice, spring water, delivered ice, and evaporative cooling. Some concern food; others concern drinks or ice itself. We also found preservation practices alongside cooling. Place, season, water, labor, money, and control of a resource change the answer. A royal example cannot stand for everyone, and a modern experiment cannot date an ancient practice.
This is an interpretation of the selected sources, not a claim that every region or household has been covered.
Modeled learning observation: “I began by looking for the old version of a refrigerator. I ended up asking who could get cooling, what they used it for, and what the evidence leaves out.” This models a possible reflection; it is not collected learner feedback.
Choose one next question, or stop with the comparison:
- What evidence survives for ordinary households in one ancient place? Our ancient court example leaves that unresolved.
- Who maintained and controlled the water or ice supply in one named community?
- What happened in a humid climate or a place without winter ice? Find a documented case before proposing an answer.
- Can we trace one city’s ice from source to household, including the workers and buyers?
- How do measured clay-pot results change with weather and design? The 2020 experimental/modeling study offers a research starting point.
Use this as an opening demonstration
Ask the opening question; run one live prompt; let the group choose a branch; open one source; revise one sentence; finish with the learners’ next question. Use Mari, the Iranian ice house, and Hopewell if you want a short contrast across ancient time, desert conditions, and unequal access. Save the other rows for follow-up.
The comparison board and revised answer are the completed written artifacts. No slide deck, generated historical image, video, or food experiment is needed. A longer class can make its own board using different places and sources.
Source and review notes
Links were checked during preparation on 2026-09-10. Historical claims above come from a specialist synthesis, heritage-site interpretation, and a museum account; these are not all primary historical records. The Mari interpretation is accessed through Jørgensen, not a direct reading of its ancient texts. MIT’s linked work reports modern research. Treat each source according to what it can establish.
Before a class, open the chosen links and prepare readable excerpts with attribution. The Iranica PDF is linked because its HTML page did not load reliably during preparation. On 2026-09-10, the Smithsonian text was available through search retrieval when automated opening failed. On 2026-09-11, the page opened in a normal browser. That records how the source was accessed; it is not a claim that learners cannot open it. The Mari descriptions above were also rechecked on 2026-09-11.
Editorial and learner-pilot review remain open, including pacing, source readability, regional coverage, and how well participants can distinguish evidence from a plausible summary. No personal learner data, copied historical images, or recordings are included.
Source: labs/it-starts-with-a-question/gallery/refrigeration-before-refrigerators.md