I Changed Two Raw Materials at the Same Time and Now the Candle Fails — How Do I Identify Which Change Caused the Problem?

Why you cannot tell from the failed candle

You changed two raw materials at the same time — for this page's worked example, the wax and the wick — and the new candle fails where the old one passed. Perhaps it mushrooms and soots, perhaps it tunnels, perhaps the jar runs hotter. The natural instinct is to look hard at the failed candle and reason out the cause. That rarely works. Both changes can produce the same symptom: a new wax can change how fast the wick draws fuel, and a new wick changes how fast it draws fuel directly. The candle cannot tell you which one did it, because it contains both.

There is also a third possibility most makers miss: neither change did it alone. Some combinations fail even though each change would have passed on its own — a wax and a wick that each work in your old system but not together. In testing language that is an interaction, and it is the reason simply reverting one change and seeing the candle improve does not prove that change was the cause. The only reliable way through is to rebuild the combinations deliberately and burn them side by side.

Before the 2 × 2: two checks that save a week

Two things come first. The first is to reproduce the failure. One failed batch can be a bad pour, a draught during the burn or a wick knocked off-centre. The D cell of the design — both changes together — reproduces the failing system under controlled conditions. If D passes when you pour it again carefully, the problem is not your two materials and the rest of the design will say so. The second is to look for a silent third change. Two changes you know about are often accompanied by one you do not: a new order of the same oil, a new carton of jars, a new scale.

Anything that answers yes goes on the change record as a third change. If you find one, freeze it: use the same oil bottle, the same jar carton and the same process in every cell of the diagnostic test, so it cannot confuse the result. CSI does not publish lot numbers or lot-to-lot consistency data for any material, which is why your own order dates and retained samples matter here. The reproducible-records post sets out what a batch record should hold so this checklist takes minutes.

The 2 × 2 diagnostic design

The design has four cells. A is the old approved candle — old wax, old wick — and it is your control. B changes only the wax. C changes only the wick. D changes both, and should reproduce the failure. Every cell uses the same oil at the same load, the same jars from the same carton, the same fill weight and the same process, poured on the same day. Three candles per cell is a working practice — enough to tell a real pattern from one odd candle — so the set is 12 candles.

Why not the quicker route of reverting one change and seeing whether the candle recovers? Because it answers a narrower question and can mislead. Reverting the wick (giving you cell B) and seeing a pass tells you the wick was involved — but not whether the new wick is bad on its own, or only bad with the new wax. To know that you would then need cell C, and to be sure your baseline has not moved you would need A. Done one after the other that is two rounds of 6 candles (12 in all, over two burn periods) instead of one round of 12; done together, all four cells burn in the same room in the same week, which removes the time difference as a source of confusion. The one-variable-at-a-time post explains when the sequential route is still the right one.

Reading the result

The interaction row deserves attention because it is the one that surprises people. If B and C both pass and only D fails, neither material is bad: the new wax works with the old wick and the new wick works with the old wax. What does not work is that particular pair. The practical choice is to keep one of the two changes — whichever you needed more — and go back to the old version of the other, with the passing cell's candles as your evidence. If you truly need both changes, the new pair is a new candle system and must be developed as one, starting with a wick test in the new wax.

For lines you record as numbers — consumption in grams per hour, time to full pool in minutes, a 1–5 throw score — the 2 × 2 also lets you estimate the size of each effect. Average the three candles in each cell. The wax effect is the average of the two new-wax cells minus the average of the two old-wax cells: (B + D − A − C) ÷ 2. The wick effect is (C + D − A − B) ÷ 2. The interaction is (D − B − C + A) ÷ 2 — how much more D differs than the two single changes would predict. You do not need statistics software; you need a clean sheet and the same line measured the same way in all twelve candles.

The priced diagnostic kit

The worked example: a 200 g candle in a White Matte Glass Jar, Cinnamon Vanilla Fragrance Oil at 8%, where the maker moved to Premium Coconut Soy Wax CSI 468 and Cotton Braided Wicks Ready Made at the same time. The old wax and old wicks come from your stock or retained samples (1,104 g of wax and 6 wicks for cells A and C). The oil comes from one bottle in your stock: 192 g for all 12 candles, about ₹928.90 at Cinnamon Vanilla's 100 g rate. The kit buys the new materials and fresh jars. A 5% pour-loss allowance on the wax is a labelled working assumption.

₹2,157.00. Most of it is jars; if you would rather not use new jars for a diagnostic, use jars from the same carton as the failing batch so the jar is frozen too. CSI 468 is ₹650.00 a kilo at 500 g and 1 kg and ₹637.00 a kilo at 5 and 10 kg — a 2% step that only matters for production — so 1 × 1 kg + 1 × 500 g is simply the cheapest cover for 1,159.2 g. No fragrance ceiling is published in this data for CSI 468; check the product page or ask CSI on WhatsApp, and keep the load identical across all four cells regardless.

After the diagnosis

The 2 × 2 tells you the cause; it does not approve anything. Whatever you keep — the new wax with the old wick, the old wax with the new wick, or a newly developed pair — goes through the normal single-change release: the full burn to the end of usable wax, the storage checks and the decision rule. The diagnostic candles are a head start, because the cell you keep has already burned side by side against a control; take one candle of that cell to the end of life and pour a storage pair if the change touched storage lines. The wax-supplier post and the wick post list those lines for each component.

The last row is the most valuable. Most two-change failures happen because two changes arrived at the same time for different reasons — a wax price rise and a wick stock-out in the same month. A standing rule that only one burning-system material changes per release, with the second held until the first is approved, prevents most of them. The change-control hub builds that rule into its tiers, and the change-control form gives it a place on paper.

Sometimes two changes truly cannot be separated in time — both suppliers are out of stock at once. That is a different problem from diagnosing a failure after the fact, and it has its own design: the simultaneous wax-and-fragrance post plans a 2 × 2 with a control before anything fails. And if you would like to avoid the situation entirely, the procurement post and the approved-supplier list stop purchasing from making the second change for you. For a wax change on its own, the wax-change checklist lists every line to retest.

Frequently asked questions