My Production Batch Has the Correct Total Amount of Fragrance Oil but Individual Candles Smell Different — How Can the Overall Formula Be Correct While Distribution Is Inconsistent?

How a correct total can hide uneven candles

A fragrance percentage is an average. It describes the whole melter, not any single candle poured from it. If the oil is blended perfectly, every jar carries the average. If it is not, some jars carry more and some less, and the batch total is still exactly right.

Take a 20 kg Bergamot Amber batch at 8% of total weight: 18.4 kg of wax and 1.6 kg of oil, filling 100 candles of 200 g. Each candle should hold 16 g of fragrance. Now suppose half the candles came out at 14 g and half at 18 g.

Both batches pass a weighing audit. Only one of them produces 100 candles that smell alike. That is the whole answer to the question: the formula governs the pot, distribution governs the jar, and they are checked in different ways.

Bergamot Amber makes the point sharply. Its product page describes how the character moves with load: at 6% the bergamot leads, at 8% it is balanced, at 10% the amber comes forward and the vanilla and musk base becomes more prominent. So a 7% jar and a 9% jar from the same batch may not just differ in strength. A customer could describe them as two different scents.

Where unevenness gets into a correctly dosed batch

Uneven candles from a correct total usually come from how the oil met the wax and how the wax left the vessel. The mechanisms that act while scented wax waits are covered on the companion page about first and last candles, so here the focus is on what happens at the moment of addition and transfer.

Several faults that are nothing to do with fragrance can look exactly like uneven dosing. Rule them out before you rebuild your mixing method.

Measuring the spread without a laboratory

A small maker cannot assay fragrance content candle by candle, but you can still get an honest picture. The trick is to make each candle traceable, then compare them blind.

Number every jar in pour order and write the pitcher number beside it. After the stated minimum cure on the oil's product page, 48–72 hours for soy in Bergamot Amber's case, pick candles spread across the run. Ten from a 100-candle batch, roughly every tenth jar, is a suggested starting point to validate. Weigh each one to rule out fill weight, code the lids, and have a second person rank cold throw without knowing the codes.

For hot throw, burn candles one at a time in the same closed room for the same time, never leaving a burning candle unattended, and stop any test that smokes or flares. If the rankings line up with pour position or pitcher number, you have a distribution problem. If they scatter randomly and fill weights vary, look at filling rather than mixing.

What to do with 100 candles that smell different

Hold the batch. Then use your position data rather than your nose alone to decide what goes where.

If the pattern is clear, for example the first two pitchers are weak and everything else is consistent, you can segregate those jars and release the rest after a confirming burn test. If there is no clear pattern, treat the whole batch as unreleased.

There is one piece of good news hidden in the maths. Because the total was correct, remelting the candles together and blending thoroughly pulls them back towards the average. Twenty jars at 14 g plus twenty at 18 g hold 280 g plus 360 g of oil, which is 640 g across 40 jars, or 16 g each: 8% again. The risk is remelting only the weak jars, which just produces more weak candles. Remelt in a mix that reflects the whole spread, add no extra oil, and keep reheating below the oil's published flashpoint and off the melter's maximum setting.

Controls that keep every jar close to the average

Distribution is a process property, so the controls are process steps written into the batch sheet, not changes to the recipe.

Keep the blend, transfer and fill rules identical from batch to batch, and change one thing at a time when you try to improve them. If you do decide to test a longer blend or a different stirring tool, run it on a 5 kg batch with the same traceable sampling before trusting it at 20 kg. Do not invest in mixing equipment until a controlled comparison has shown that mixing is the constraint rather than filling, jars or cure age.

Where this page fits

CSI already has guides that cover parts of this question. This page covers the failure investigation for the question above; these go deeper on the rest:

Three things to buy for this job

Frequently asked questions