The number your rule guarantees isn't always in the rule
A tenant in Beijing has a legal guarantee about how warm the flat will be this winter. It is not written in the Beijing rule. That rule, in its twelfth article, sets the dates of the heating season and then requires the supplier to keep bedrooms and living rooms at "the temperature requirements of the current national residential design code" — a pointer, not a number.
I spent a morning comparing four Chinese heating regulations: Beijing's municipal measures, the Heilongjiang and Shandong provincial regulations, and Xi'an's city regulation. The comparison itself is the interesting part, because the four documents answer the same question four different ways, and three of the differences are things you can only see by putting them side by side.
Three shapes of the same promise
- Heilongjiang names 20°C, all day, in bedrooms and living rooms, and then describes the measurement point: halfway into the room from the door, 1.4 metres up (art. 32).
- Shandong names 18°C and attaches three preconditions — outdoor temperature at or above the system's design minimum, an envelope that met the design standard when built, and a heating system running normally (art. 21).
- Xi'an names 18°C, and leaves "other parts" of the flat to the national design code.
- Beijing names nothing. The number is 18°C, and it comes from two standards outside the regulation, with a condition attached that the regulation also does not mention.
If you are writing code against these rules, or arguing with a landlord, the shape matters more than the value. A rule that names a number gives you one lookup. A rule that points at a standard gives you a chain: regulation → design code → the clause that actually sets the value → the condition under which the clause applies.
The escape clause is in the government's FAQ, not the rule
Beijing's district government publishes a heating FAQ. It states the 18°C figure and names the two standards behind it (GB 50096-2011 and GB 50736-2012). The same page then explains that the number holds when the outdoor mean temperature is above −7.6°C. Below that, the city says it requires suppliers to run their systems at maximum design load, and that it does not require the indoor temperature to be met.
That is a real limit on the guarantee, and it appears in a Q&A page rather than in the regulation. This is the sort of thing I now check for by default: after finding the clause that states the obligation, I look for the page that explains what the obligation does not cover. Enforcement guidance tends to live there, not in the statute.
Arithmetic I had to do myself
The rules give dates, not lengths, so the lengths are my arithmetic, counting both the first and last day:
| City | Season | Days |
|---|---|---|
| Beijing, Xi'an | 15 November – 15 March | 121 |
| Harbin | 20 October – 20 April | 183 |
| Harbin this winter | opened 10 October; end unconfirmed | 193 if it ends 20 April |
Nothing in the rules says "183 days". Harbin's own July consultation document does, which gave me a check on the counting convention: my inclusive count reproduced their number exactly. When you compute a figure a document does not state, find one document that does state it nearby, and use it to calibrate the convention. Off-by-one in a season length is exactly the kind of error nobody catches.
Version control is the hard part
Chinese municipal websites often republish provincial texts. The Heilongjiang regulation I read is hosted on a district government site, uploaded in 2023; the Shandong text lives on a county government site and matches the March 2025 amendment quoted by the provincial people's congress. Neither is the canonical publisher, and neither page makes it obvious which amendment it contains.
Two consequences worth writing down:
- Search snippets of a regulation can be two versions old. An early search result showed Heilongjiang's floor at 18°C with a different article number; the current text says 20°C at article 32. I trusted the full text with an amendment date over any snippet.
- A news report of adoption is not the final text. Xi'an's government page reports the regulation being passed in December 2017 and summarises the refund tiers. I used it for the tiers and said so in the piece, rather than pretending I had the article text.
What a comparison like this is good for
It produces a table nobody publishes: floor temperature, season length and refund schedule for four places, each cell traceable to an article number. It also produces honest gaps. I could confirm Harbin opened its season ten days early this year; I could not find the published final text of the decision that may have moved it, so the article says the length is conditional. Where the money goes after a failed season (Heilongjiang has a 50%/100% daily refund table in the regulation itself; Xi'an has 20%/50%/100% bands; Beijing routes disputes to third-party measurement and then to the regulator or a court) is the part I would want if I were renting.
The full comparison, with article numbers and the source for every figure, is on my site: China's heating guarantee: 18°C, 20°C, or no number stated.
Drafted with an AI assistant from my own reading of the primary documents; every clause number and link here was checked against the source.
原文:https://www.a8q9.cn/en/article/the-number-your-rule-guarantees-isn-t-always-in-the-rule-3708a6.html