You pour 30 candles from the same pot, same wick, same wax. The label says 42 hours. In our QC burn room, three of them die at 34. The other 27 make it past 44. That 10-hour spread in a single batch is not bad luck. It is a spec violation you can catch before the pallet leaves the factory.
The Spread Is the Signal, Not the Average
The number your buyer sees — “42-hour burn time” — is a mean, and a mean hides the two candles at each end. In our factory, we burn three candles from every 50-unit batch and log the spread: the difference between longest and shortest, expressed as a percentage of the mean. A tight batch stays within 8%. Once that number crosses 10%, one of three things drifted during the pour: temperature, wick size, or fragrance load.
Soy wax pours cleanly between 60 and 75°C for most 464 and 444 blends. But that is the pot temperature at the moment you tip the ladle, and in a 50-jar run the first pour is 4 to 6°C warmer than the last unless you hold the melt tank at 65°C with a circulating heater. A 5°C shift across the batch changes solidification rate. The early jars set faster, bond the wick more tightly, and burn 3 to 5% shorter. The late jars do the opposite. That is where your 34-hour and 46-hour candles both come from.
Three Numbers, One Spreadsheet Column
We log three values per batch and we do not ship until all three are in band. Pour temperature range: the gap between the first and last ladle in the batch. Target is 4°C or less. We hold the tank at 64°C with a circulating heater, and a 30-minute pour of 25 jars stays within 3°C. Wick spool ID: the specific spool number and the measured wick diameter at both the head and the tail. A 0.1mm drift across a 2,000-piece spool of paper-wrapped wick is normal. A 0.2mm drift is not, and it costs you 8 to 12% of burn time. Fragrance oil weight per jar: we weigh every jar before and after addition. Target is ±0.3g per 226g jar, which is ±0.15% FO. A 9% load instead of 7% changes burn time by roughly 4% because the oil burns at a different rate than the wax matrix.
These three numbers go on a one-line batch card that rides with the carton. If a buyer’s customer calls in a “dies early” complaint, we pull the card and tell them which jar was in the batch, what the pour temp was, and which wick spool it came from. That is the difference between “we will send a replacement” and “here is exactly why that one ran short.”
The 2-Hour Rule: When to Scrap a Batch
Here is the line we hold ourselves to: if the burn-time spread across a sample of five exceeds 20% of the mean, the batch goes to rework. A 40-hour mean with a 32-hour candle in the sample is a 20% gap. We re-melt, re-pour, and the original wax goes into our 9b batch for pillars and wax melts, where burn time is a looser spec.
In one run last spring, we pulled 140 jars because the wick spool had a 0.2mm diameter shift in the last 400 pieces. The customer, a German importer who tests every third carton, found two candles in their 40-hour range at 31 hours. We caught the next spool before it hit the pour line. The scrap cost was about €220 in wax and wick. The credit-avoidance was worth ten times that. The burn-time spread is the one number that ties pour temp, wick geometry, and fragrance load into a single, customer-visible metric. You do not need a lab. You need a stopwatch, three jars per batch, and a spreadsheet column you actually fill in.