My Candle Smells Different at the Bottom of the Jar than It Did at the Top — Could the Fragrance Oil Be Distributed Unevenly Through the Wax?

Does oil settle in a set candle?

Fragrance oil in a container candle is held in the wax as it cools and sets. Once the candle is solid, we know of no evidence that the oil migrates downward through it, and CSI does not claim otherwise for any of its waxes. Oil can move to the surface, which is what sweating is, and that has its own cluster in this series, starting at the droplets-on-day-three hub. What is plausible is that the oil was never evenly distributed to begin with, and that happens in liquid wax, not solid.

In the pot, oil that is added to wax that has cooled too far, or stirred for a few seconds, may not incorporate fully; it can sit as a film or collect unevenly before the pour (a tendency, not a certainty). The Eco Soy CSI 400 page is the only CSI wax page that states a stir time, "stir gently for 2 minutes"; for other waxes, a thorough stir is judgement, and the stated add or pour temperature on the wax page governs. The detailed mechanics of adding oil too cool are in the added-too-cool post, and of a 20-second stir in the 20-second stir test.

Four reasons the bottom smells different

The table ranks the likely causes in the order we would suspect them. That order is editorial judgement, not measured data; CSI publishes no failure statistics. The right-hand pair says which slice comparison rules each one in or out.

Cause 1 is the likeliest in our judgement because it needs nothing to have gone wrong: every burn melts the upper wax again, lighter notes tend to leave a hot pool first, and by the bottom of the jar the remaining wax has been through the most heat. With an oil like Dark Vanilla, described as "rich vanilla and spice opening into deep caramel and cream, settling into dark amber and warm musk", the spice and vanilla top can recede and leave the caramel and amber louder; that reads as "different", not "weaker". Repeated heating producing an unpleasant note is its own diagnosis in the several-burns drift post.

Cause 2 is easy to miss because it is not about depth at all. The top was smelled when the candle was young; the bottom weeks later. A candle's character can shift with cure time, which the day-1 versus week-2 descriptor log tracks. The twin is the same age as the burned candle, so its slices tell you what the wax smells like today.

The sacrificial-slice test

The test compares wax, not candles. It uses an unburned twin from the same pour as the candle that smells different: same pot, same session, ideally poured next to it in the sequence. Being unburned, its bottom has never been heated since it set, and being the same age, it removes cure from the comparison. If you have no unburned twin, the test cannot answer the question for that batch; pouring a new pair checks your process today, which is still useful, but it is a different question.

What the test costs

The main cost is the twin. In the worked example (Dark Vanilla at 8% in CSI 464 in the 150 g jar with a C1 wick), sacrificing it consumes ₹94.80 of oil (12 g at the 100 g price), ₹70.02 of wax and ₹5.90 of wick, ₹170.72 in all; the jar (₹94.40) is cleaned and kept. Costs exclude labour, freight, packaging and GST. Tools: a burner, a scraper and a scale from CSI total ₹1,121.00 if you own none of them, and nothing if you use saucers over hot water, a spoon and your kitchen scale. If you have no twin and pour a fresh process-check pair instead, that pair is ₹530.24 of materials, and it tests your method today rather than the batch in question.

Reading the slices

Every outcome in that table can be settled without buying anything; only the first might call for stirrers or a thermometer, and only if you have none. That is not a sales line; it reflects what the test usually separates. When the slices match, you have learned that your mixing was fine, which is worth knowing before you change a process that works. When they differ, you have learned something more valuable: the problem is upstream of every candle in that batch, and fixing it costs almost nothing.

A worked reading

Say the burned candle's bottom smelled of caramel and amber where its top had been vanilla and spice. On the slice test, both judges call T and B "same" in both orders, and describe both as "vanilla, spice, some caramel". R is called "different" from B in both orders and described as "caramel, amber, a bit flat". That is the second row: the wax went in evenly, and the burned candle's heat history moved it from the top of its note pyramid to the bottom. The oil is doing what its notes describe, and your mixing passed. These descriptions are illustrative; yours are the data.

If the slices differ: fixing the pour

A real T-versus-B difference means the jar was filled with wax that was not uniform. The causes are all upstream and all process: oil added below the wax maker's stated temperature, a short stir, a pot left standing between pitchers, a pitcher drawn from the top of an unstirred pot, or a top-up poured from different wax. None of them is fixed by a different oil.

After changing the process, pour a pair and repeat the slice test on one of them once it has cured, before you trust the next batch. For CSI 464 in the 150 g jar, pour at 80–85°C: the jar page's "Safe for hot pour up to 85°C" and the wax page's 80–90°C overlap there. A thermometer is the one tool this fix needs if you pour by feel; the Pen Thermometer is ₹354.00, and its page states no range or accuracy, so check it against a second thermometer you trust before relying on it. Separation diagnosed at the level of the whole soy system, as opposed to one jar, is the separation hub.

Before you spend anything

We sell the oil, burners, scraper, scale and stirrers on this page. Here is where we would stop you.

For the wider question of which variable to change first when a candle underperforms, the cluster hub orders the checks and the which-change-first ladder costs the levers. If this candle also smells burnt at the flame, run the smell-source isolation first; a burnt note is a combustion question before it is a distribution one.

Tools and the worked-example oil

Five items in test order: the flameless warmer, the scraper, the worked-example oil, the scale for equal slices, and the stirrers for the process fix. Chosen on what each page states, price per unit and stock on 26 September 2026; CSI has no measured throw data, so nothing is ranked on throw. The oil table shows every size, 150 g candles at 12 g, and the cost of 12 g at each size.

Frequently asked questions