I Want to Fix My Candle Without Wasting Dozens of Test Batches — How Do I Troubleshoot One Variable at a Time Across Wax, Wick, Fragrance Load and Temperature?

Where the wasted batches actually go

Makers who say they have poured "dozens" of test batches rarely lack effort. What usually went wrong is that each batch changed two things, or had no candle poured alongside it to compare with, or was one candle judged on one evening. Those batches cost the same as good ones and answer nothing, so the next batch starts from the same uncertainty. Before you plan anything new, look back at your last few tests through the table below. It ranks the patterns that, in our judgement, most commonly explain wasted testing, most common first; the ranking is our judgement from how these problems tend to present, not a measured frequency.

Rows 1 to 3 are the reasons this planner insists on a same-day control and duplicates. Row 4 is why every round is judged at a fixed cure age. Row 5 is why round 0, before any pour, is a weighing check: if your 150 g candle at 8% is meant to hold 12 g of oil and 138 g of wax (load as a percentage of total weight, CSI's convention), weigh one pour and confirm it. Row 6 is a stopping rule, below. If you do not yet know which symptom you are chasing, the evidence burn in the cold-versus-hot triage hub tells you where to start, and the order in which to attack several problems at once is the five-problem dependency map; this page assumes you know what you are trying to fix and want to spend as little as possible proving the fix.

The round planner

Why not test everything at once? You can, and a full grid shows how every variable interacts with every other. It is also expensive. Two waxes × three wicks × three loads × two pour temperatures is 36 recipes, and 72 candles in duplicate. The planner below gets to a working candle with far fewer by using what candle makers already know about the order of dependence: a wax change moves the wick, the load ceiling and the temperatures; the load moves the wick; the wick rarely changes what the wax can hold. So settle the variables that change the others first, and hold everything else still.

Three design choices carry the whole method. Duplicates, not triplicates. Two candles per arm catch the single rogue candle (an off-centre wick, an air pocket) at the lowest cost; a third candle is bought only when the two disagree (a stopping rule below), which should be the exception when the pour is careful (our judgement). When you need to rank several loads with confidence rather than pick a winner, the triplicate grid in the 8%-beats-10% paradox is the stronger design. The control is poured, not remembered. A candle you poured three weeks ago has a different cure age and has lived in a different room; a pair poured today beside the arms has neither problem. Also keep one unburned candle of your original recipe from round 1, lidded, as an anchor you can look at if anything seems to drift. The winner becomes the control. Each round starts from the best candle so far, so the comparisons stay one step apart.

Two more notes on the arms. The load ladder stops at 9% of total weight because CSI 464's page states a ceiling of 10% without saying whether it means total weight or wax weight; read on wax weight, 10% is 9.09% of total, so 9% is the highest rung that respects both readings, and Citrus Lemon's page also says "candles (6-10%)". Never add a rung above the lower of the two ceilings. And round 3's temperatures both sit inside the figures the pages give: CSI 464's page states "Pour Temperature 80°C – 90°C" and the jar page says "Safe for hot pour up to 85°C", so 80 to 85°C is the window in this jar. We do not invent a better temperature inside it; the round finds out whether it matters for your candle at all.

What each round costs, and what to buy for it

Each candle in the worked example costs ₹209.86 in materials: ₹36.00 of Citrus Lemon (12 g at the 100 g price of ₹3.00/g), ₹70.02 of CSI 464 (138 g at ₹0.51 per gram), ₹94.40 for the jar and ₹9.44 for a C2 wick. The jar is 45.1% of round 1. Costs exclude labour, freight, packaging and GST throughout.

Read the table three ways. Planned: rounds 1, 2, 3 and 5 are ₹4,180.71; round 4 is a contingency you set aside but hope not to spend. Best case: if round 1's winner fixes the problem and round 5 confirms it, you stop at ₹2,089.57, which is 13.5% of the grid. Worst case: all five rounds, ₹5,495.56, are 35.6% of the grid, or ₹9,957.91 less. The grid is not wrong; it answers more questions than you have. For one misbehaving candle you rarely need to know how the 6% load behaves in a wax you are not going to use.

Buy round by round

Because each round's result decides whether the next one happens, buy for the round in front of you. Wax is the easy case: CSI 464 is ₹0.51 per gram at 1 kg, 5 kg and 10 kg alike, so a 1 kg bag per round costs the same per gram as a 5 kg bag, and all five rounds need four 1 kg bags (₹2,029.60) if you buy as you go. Two 500 g bags would cost ₹12.60 more than one 1 kg. Oil is the case where the per-gram price misleads: the five rounds use 309 g, so a 500 g bottle (₹1,420.00) leaves 191 g over, while four 100 g bottles bought as needed (₹1,200.00) leave 91 g and cost ₹220.00 less cash. If round 1 solves it, you need only two bottles (₹600.00). Jars: 26 across all five rounds, so a 20-pack and a 10-pack (₹2,832.00), or fewer if you reuse clean, undamaged jars from finished rounds. Wicks: 18 C2 (a pack of 20), four C1 and four braided.

Stopping rules

Every round ends in one of three outcomes, and the decision rule for each is written before you pour. The rule compares the arms with the control pair on the round's primary metric (for round 1, the burn: full pool time, pool depth, flame, soot and jar heat read with a non-contact thermometer, never by touch; for round 2, a blind throw rank):

Then the programme-level rules, which decide whether there is a next round at all:

A word on time, because it is the real price of this method. Each round needs a cure before it can be judged (Citrus Lemon's page says to cure soy candles "48-72 hours before burn testing"; many makers judge throw at day 7 or later), plus several burn sessions. Call it roughly ten days a round as a worked example, so the full programme runs for weeks where the grid would run once. That is the trade: fewer candles for more calendar. If you have a hard date, the fix is to start round 1 sooner, not to merge rounds.

Running one round from start to finish

Round 1, the wick round, as the worked example. Six candles from one pot of CSI 464 with Citrus Lemon at 8% of total weight: 12 g of oil and 138 g of wax each, 72 g and 828 g in all.

Rounds 2 and 3 follow the same six steps with one difference: throw is judged blind. Code the jars, have someone else light them in random order, and rank them from a fixed spot with fresh air between rooms; the full protocol is the blind-scoring method. Keep the burn guardrails from round 1 as pass/fail checks in every later round, so a load that smells stronger but drowns the wick does not win. If your round 1 arms were chosen from the wrong starting sizes, the way to pick three sensible rungs for a jar is the first-three-wicks method.

What not to buy while you test

We sell every item in this plan. The cheapest outcome for you is that round 1 solves it, you confirm with four candles, and you have spent ₹2,089.57 on materials instead of several times that on test batches with no controls. That outcome also means you buy less from us this month, and we would still rather you had it.

The materials the planner is costed on

Five items: the constant wax, the jar, the two wick lines and the worked-example oil. Chosen on page statements, price per unit and stock on 26 September 2026; CSI publishes no burn or throw data, so nothing here is ranked on performance. A Candle Making Weighing Scale (₹354.00) and a Pen Thermometer (₹354.00) are round 0 if you do not already own them; their pages state no capacity, resolution or accuracy, so check them against a known weight and against each other.

Frequently asked questions