The Colour of My Fragrance Oil Changed Between Batches but the Supplier Says It Is the Same Product — What Should I Test Before Making Candles?

What the bottle can and cannot tell you

When two bottles of the "same" oil look different, three explanations are open, and the colour alone does not choose between them. The lot varies. Fragrance oils are blends of many materials, some of them natural extracts whose colour can differ from one production run to the next. A supplier can truthfully call two lots the same product while one is paler or more golden. The old sample has changed. Oils commonly darken with age, air in a half-empty bottle, warmth and light. The retained bottle you are comparing against may not look the way it did on arrival. The composition has changed. That is possible, but it is the one explanation colour can't confirm. It needs the scent and burn checks that the smell-different go/no-go gate scales to the size of your run.

CSI publishes no lot or batch release data, composition or reformulation notices for its oils, and we will not pretend to read a lot from a photograph. What we can do is give you the lot or batch numbers of both orders on WhatsApp so your records line up, and send MSDS and IFRA documentation on request. The rest is a test you run in your own wax, because that is where a customer sees the colour.

One more reason not to judge from the bottle: dilution. At 8% of total weight a 150 g candle holds 12 g of oil in 138 g of wax. A faint golden tint in the bottle often disappears into a creamy soy. A deep amber one may not, and neither may the slow darkening some components show over days in light. Both of those are in-wax, over-time effects, which is why the test runs for fourteen days rather than an afternoon.

Colour patterns, likely causes, and what settles each

The table lists the patterns makers report after a lot change, in the order we would suspect their causes, most common first. That ranking is editorial judgement: CSI has no discolouration test data and publishes no lot data. The two right-hand columns are the useful part. Each is an observation from the test below that rules a cause in or out.

Patterns 2 and 6 are why a like-for-like comparison is the whole point of the test. An old bottle, or an old candle, has had more time to change, so it can make a normal lot look pale or a pale lot look normal. Pour both lots on the same day, at the same temperature, into the same wax, and age them side by side.

When the colour of the oil actually matters

A darker lot is not a safety or performance fault in itself. What it threatens is a look you sell, and your customers' trust when the second order doesn't match the first. How much is at stake depends on your range, not on the oil. Use the table to decide how much testing the lot deserves before production.

The oil you use sets the baseline. Vanilla Bean's page states "Vanillin Content: 2.5% (Potential discoloration; test with lighter waxes…)", Dark Vanilla's says "Vanillin present - note for light-coloured wax", Gardenia's says "Vanillin content 0.2% - minimal discoloration risk", and Eucalyptus Spearmint's says "No vanillin - zero risk of discoloration". Vanillin and some other aromatic components commonly cause yellowing or browning over time, faster in light. So a vanilla in a white clear jar will tend to darken whatever the lot, and a lot comparison there is about how fast, not whether. Which component of a custom blend is doing it is a different test, the isolation set in the discolouring-component post.

The in-wax colour test, step by step

The design has three arms and two storages. O is the old lot, N the new lot, U an unscented control from the same wax. Light and dark storage for each makes six candles. U is what makes the test readable: if U changes, the wax or the light moved, and you subtract that before blaming either lot. These candles are never burned, so they need no wicks.

Costs are worked examples and exclude labour, freight, packaging and GST. With no old oil left, drop O: four candles (N-L, N-D, U-L, U-D), 24 g of oil and 576 g of wax, ₹859.46 in materials. You then compare against dated photos of the old lot, which is weaker evidence. Now weigh the test against what it protects. Worked example: one production candle of a white vanilla line costs ₹262.07 in materials (₹91.75 of oil at the 1 kg price, ₹70.02 of wax, the jar and a ₹5.90 C1 wick). A 200-candle run is ₹52,414.64, and uses 2,400 g of oil. The whole colour test is 2.6% of that run.

Reading day 14: the decision rule

Two limits keep the result honest. First, the test answers colour, not scent. A new lot that passes on colour still goes through the scent gate before a large run, and a lot that fails on colour may be perfectly good for your tinted range. Second, the test speaks for your wax and your light only. A lot that passes in creamy CSI 464 on a shaded shelf may show more in a whiter wax under shop lights. Test in the conditions you sell in.

What not to buy while the question is open

A colour worry makes people reach for additives and replacement oils. Most of those purchases don't answer the question, including some of ours.

If the new lot does fail, you have three real options, in order of cost. Re-route it to a tinted, amber-jar or deep-dyed line where the difference doesn't show, which costs nothing. Change the white line's oil to one whose page states little or no vanillin, and test it the same way. Eucalyptus Spearmint at 12 g a candle costs ₹36.82 at the 1 kg price against ₹91.75 for Vanilla Bean, so the switch saves money per candle, but it is a different scent and your customers will notice. Change the vessel for that scent, which changes the product. None of the three needs an additive.

Oils, wax and jars for the test and the fix

Five items: three oils that span the vanillin statements on CSI pages (stated 2.5%, stated 0.2%, stated none), the uncoloured wax the test runs in, and the clear jar that shows colour through the wall. They are chosen on page statements, price per gram and stock on 26 September 2026; CSI has no measured colour data, so nothing is ranked on how white it stays. Oil tables show every size, 150 g candles at 12 g, and 24 g colour-test pairs.

Frequently asked questions