1. 01

   We do not test candles

   Ember & Main runs no laboratory and burns nothing. Every burn time on this site was published by somebody else, and we say who. If we ever start testing, it will appear as a separate kind of claim with its own protocol, and it will never be blurred together with figures we merely collected.

2. 02

   A figure needs a source and a date

   Nothing is recorded unless someone opened the page it appeared on and read it. We store the publisher, the page, and the date it was read. A number without those three things is a rumour, and rumours are how a comparison site quietly becomes useless.

3. 03

   The maker outranks the retailer

   Manufacturers and shops routinely publish different burn times for the same product, and the shop's figure is usually the higher one. Where they conflict we lead with the maker's number and show the others beside it, because the party with the least incentive to inflate is the one who made the thing.

4. 04

   Ranges collapse to their lower bound

   When a maker says a candle burns sixty to eighty hours, we record sixty. Cost per hour calculated from the optimistic end of a range flatters every candle that publishes one, and penalises the honest makers who publish a single conservative figure.

5. 05

   Conflicts are the interesting part

   Where publishers disagree by more than a rounding margin, we say so on the page rather than silently picking a winner. A third-of-a-candle gap between two published figures tells you something real about how the number was arrived at, and it is information nobody else collects.

6. 06

   Stated is never measured

   Every burn figure on this site is labelled as stated by its publisher. None of them are measurements, ours or anyone's, unless a page explicitly names the tester and their method. We would rather the distinction be tedious than invisible.