I Accidentally Used a Fragrance Oil I Haven't Burn-Tested in 100 Candles — Should I Test Finished Samples Before Deciding Whether the Batch Can Be Sold?

What the batch is worth and what a sample costs

Cost the batch before you argue with yourself about the sample, because the ratio is what settles it. This worked example uses a 150 g fill and a load of 8% of wax weight — the convention CSI product pages use, and the one the Luxury Soy Wax Chunks page states when it gives "8% to 13% of total wax weight". On that convention a 150 g candle divides as 138.89 g of wax and 11.11 g of oil. On the finished-candle convention the same candle would carry 12 g of oil, which is why the sheet has to say which one you used.

Two things are worth noticing in that table. The oil is the most expensive line, at ₹101.33 a candle against ₹94.40 of glass and ₹83.19 of wax — which is a useful corrective to the instinct that fragrance is the cheap part. And the jars and wicks are already spent in a way the wax and oil are not: if the batch fails, wax and oil can be remelted and recovered, while a wick set in cured wax is gone and the glass has to be cleaned out by hand.

Which four candles to burn

The sampling plan matters more than the number, because it decides what the result is evidence of. Six steps, and the first one happens before you light anything.

The five outcomes, and what each one means

A sample burn has five realistic results and they do not all mean the same thing. Sorting them properly is what stops a maker binning a saleable batch out of anxiety, or shipping an unsafe one out of optimism.

Note which way the asymmetry runs. Three of the five outcomes end in "sell", and two of those three simply require you to describe the candle accurately rather than aspirationally. The two that end in "no" are both wick outcomes, and both are about how the candle burns rather than how it smells. That is the deep reason this test is worth 4% of the batch: you are not really testing the fragrance, you are testing whether an unfamiliar fuel and your existing wick get along.

The one failure you cannot repair

It is worth stating plainly, because it shapes the whole economics of this decision: a wick set in cured wax cannot be swapped and sold as a tested candle. You can heat the top surface, you can dig, you can pour a new layer around a new wick — and you will end up with a candle whose interior geometry you cannot see and cannot vouch for. The honest routes when a wick fails are to remelt and recover the wax, or to accept the candles as non-saleable stock.

Recovery arithmetic, so the choice is concrete. If the 96 candles have to be remelted, you recover the wax and the oil — ₹8,319.44 and ₹10,133.33 in the full batch, so roughly ₹17,714.66 across 96 candles, minus whatever you lose to the remelt itself. You do not recover the 96 wicks at ₹5.90 each (₹566.40) and you spend labour cleaning 96 jars. Measure your own remelt loss rather than trusting a figure: weigh the wax going into the pot and weigh what comes out, and the difference divided by what went in is your loss percentage for that batch and that equipment. Every honest number in a remelt calculation is one you measured.

What the product page can and cannot tell you

Before you buy an oil you have not tested, it is worth knowing exactly how much a product page can substitute for a burn test. The honest answer is: it narrows the range, and it does not close it.

Compare two CSI pages. Santal Noir — the oil in this worked batch — gives a detailed olfactory profile, top, heart and base notes, and how the scent develops. What it does not state is a fragrance load, a flash point, a vanillin content or a cure time. Now look at Sandalwood Fragrance Oil, in the same broad family: it states up to 10% for soy wax with a 48–72 hour cure, up to 6% for paraffin, wax melts up to 12%, a flashpoint of 203°C and 0% vanillin with no discolouration. Two oils, two very different amounts of published information.

The two "no" rows are the reason this page exists. Wick and throw are properties of a combination — this oil, at this load, in this wax, in this jar diameter, on this wick — and no supplier can publish a figure for a combination you assembled. Published reviews help: one CSI customer describes an Apple Cinnamon candle filling a customer's home on arrival, which is real published experience with a real oil. It is still not a measurement of your jar. For a specific oil, ask us for the SDS and any IFRA documentation on WhatsApp, and then burn a candle.

What this batch was built from

Three cards covering the lines that decide whether a batch like this passes: the jar whose stated fill makes the arithmetic possible, the wick you will be re-sizing if it fails, and the thermometer that stops the next batch being untested in a different way.

The three-candle pre-test

Prevention here has a price, and it is lower than the sample you are about to burn. Before any new oil goes into a 100-candle run, pour three candles of it and nothing else.

Three candles of this recipe cost ₹854.48 of material — ₹284.83 each — against ₹1,139.31 for the four-candle sample you now have to take out of finished stock, and against ₹27,343.47 of exposure. The pre-test also costs less in a way that does not show up in rupees: it happens before you have bought 100 jars, so a failure costs you three jars rather than a hundred.

The second prevention habit is a purchasing one, and it is cheap. Most CSI oils list a 15 gm size, and where they do not, the smallest listed pack is usually 50 g or 100 g. Santal Noir lists at 15g ₹166 — enough oil for one 150 g test candle at 8% of wax weight with plenty over for smelling. Buying a 1 kg bottle of an oil you have never burned is not a saving; it is a bet with a 1 kg stake. And once the pre-test passes, write the whole recipe down against a batch code, because the thing that turns a tested recipe into an untested batch is almost always a record that was never made. The record fields are set out in the companion guide on storing a blend so the candle can be reproduced.

Frequently asked questions