How Do I Systematically Diagnose Candle Soot from Wick Size → Wick Trimming → Fragrance Load → Dye → Airflow → Burn Duration Before Replacing Raw Materials?

Your six variables, re-ordered by cost

Your question, restated: a candle soots, and before you buy new wicks, oil or dye you want a method that works through wick size, trimming, fragrance load, dye, airflow and burn duration and tells you which one is responsible. The title puts wick size first, and among the formula variables it is usually the first suspect. But testing it means new candles and ten days; testing trimming means a pair of scissors and one evening. So the protocol splits the six by what testing them costs, and orders each group by how often it explains soot (our judgement).

Two tendencies sit behind the bottom three rows, and neither is CSI data: higher fragrance loads and heavier oils (sweet, vanilla-rich, resinous, dark gourmand) commonly build more carbon on the wick, and heavy dye loading is commonly linked to more carbon too. No CSI oil or dye page gives soot data in the text we captured. That is exactly why they get tested rather than assumed. Each variable also has its own post in this cluster: trimmed and still sooting, flicker and airflow, soot only after four hours, 8% to 10% and more soot and dark colours and dye. This post is the sequence that ties them together.

Stage 1: trimming, airflow, burn duration

Take 3 candles from the problem batch, the same age and lot as the ones that sooted. Each step adds one control to the one before, so by step 3 you are burning a trimmed candle in still air and varying only the session length. Use the white card (held briefly well above the flame at hours 2 and 4) and a rim photo after cooling as your evidence in every step.

Stage 1 has one limit you should know: three candles on three evenings test the top of the jar only. A candle that soots only in its last third needs the full-life approach from the start (the late-jar soot hub). And if your soot complaint came from a customer rather than your own burn, the evening you are testing is theirs: the customer-photo questions turn their answers into stage 1 conditions.

Stage 2: wick, load and dye as parallel arms

The textbook way is sequential: test the wick, wait ten days, test the load, wait ten days, test the dye. Each step is clean, but the whole thing takes about 36 days in our worked example, and a festive batch won't wait. The parallel design keeps the discipline (each arm differs from the control in exactly one variable) and runs all three at once. The control is the key: every comparison is arm against control, never arm against arm.

Why these steps? One CSI rung, because that is the smallest wick change we can offer (more on that below). Two points of load, because a smaller step is hard to see in three candles, and 6% stays inside both the Apple Cinnamon page's "6-10% for soy wax candles" and CSI 464's "Up to 10%"; use your own oil's stated range, and never go above it. No dye at all, because it answers the yes-or-no question cleanly; if the undyed arm passes, a lower-dose rerun tells you how much colour you can keep (reduce dye or change wick first?).

The recording sheet

One line per candle per session. Keep the same columns through both stages, so a stage 1 candle and a stage 2 control can be compared directly.

Worked example: 200 g Apple Cinnamon, red, on C2

The candle: 200 g in the maker's own jar, Luxury Soy Wax CSI 464, Apple Cinnamon at 8% (16 g oil + 184 g wax), red liquid dye, one Eco C2 wick. The complaint: grey on the rim after evening burns, black after a long one.

Stage 1. Day 1, trimmed in the usual living room: card grade 2 at hour 4, wick capped again by hour 4. Not just trimming. Day 2, trimmed in still air: card grade 2 at hour 4, rim grey all round, no flicker. Not airflow. Day 3, trimmed and still, 2 hours against 4 hours: the 2-hour candle's card is clean, the 4-hour candle's is grade 2. Duration-dependent, so the candle has no margin at the four hours its label allows. On to stage 2, poured on day 4.

Stage 2 results (illustrative). Control: grade 2 by hour 4 in the top third, grade 3 in the bottom third of two candles. Fails, which is good: the soot reproduced. Wick arm (C1): grade 0–1 everywhere, pool reaching the glass by the end of each session from burn two, the flame about a centimetre lower. Passes. Load arm (6%): grade 1 at the top, grade 2 in the bottom third of all three. Fails, though it helped. Dye arm (undyed): no better than the control. Dye is not the cause.

Decision on day 14: the wick is responsible. Change the spec for this scent from C2 to C1, keep 8% and the red dye, and keep the oil and wax. The spend was ₹118.00 on wicks plus stage 2 wax. The load arm's partial improvement is worth noting: if C1 had tunnelled, a lower load on C2 would have been the next thing to test, at ₹18.97 less oil per candle (₹75.90 at 8% against ₹56.92 at 6%, at the 1 kg per-gram price). What the example did not do: buy a new oil, a new dye or a new wax, or change two things at once.

When to replace raw materials

"Replacing raw materials" can mean three different things: changing a spec (a different wick rung, a lower load, a lower dye dose, all with materials you already buy), changing a material (a different oil or dye), or changing a supplier. Work in that order. A spec change is cheap and reversible; a material change resets your scent or colour; a supplier change resets everything and needs its own retest (what to retest when changing wick supplier).

Argued against our own interest: two of the six rows involve no purchase from anyone, and a third (load) means buying less oil. The cleanest-sounding fix, a new oil "that burns cleaner", is the one this protocol reaches last, because it changes the candle you sell and no CSI oil page gives soot data to justify it. For a scent that soots on every wick and every load, the vanilla-versus-lavender post sets up the per-fragrance comparison. The whole-symptom tree beyond soot (tunnelling, drowning, deep pools) is the systematic diagnosis capstone.

The CSI buy list and the step to production

CSI's sized, tabbed wicks are three rungs: C1 → C2 → Cotton Braided. The wick arm is one rung down: Cotton Braided → C2, C2 → C1, or a C2 pair → a C1 pair. There is no sized, tabbed wick between C1 and C2 or between C2 and Cotton Braided, no smaller sized, tabbed wick than C1, and one wooden-wick listing with no widths. If the wick arm passes on soot but its pool falls short, the finer adjustment is the load arm, or ask us on WhatsApp; we'll say plainly if we have nothing in between.

Argued against our own interest: if stage 1 ends it, you spend nothing. Because C1 (₹5.90), C2 (₹9.44) and Cotton Braided (₹12.00) cost the same per wick at every pack size, a bigger pack buys stock, not savings, and the C2 → C1 step saves ₹3.54 a candle. Apple Cinnamon is ₹4.74 a gram from 50 g to 1 kg, so its 500 g (₹2,371.80) saves nothing per gram over two 100 g packs either.

The step to production quantity

Production wicks = candles × wicks per candle × (1 + a defect allowance). Worked example, labelled: 200 single-wick candles a month of this scent with a 2% allowance is 204 wicks; on C1, 2 × 100 + 20 = 220 for ₹1,298.00. At 8% those 200 candles take 3,200 g of oil a month; for Apple Cinnamon, 3 × 1 kg + 500 g cover it for ₹16,602.60. Write the new spec, the arm results and the date on your spec sheet, keep two retains from the first production batch, and treat any later wick, oil or dye lot change as a new stage 2 comparison against a retained control. Reorder point = daily use × your lead time + safety stock; CSI publishes no lead times, so ask on WhatsApp.

What to buy, and what to skip

Four items for the worked example, chosen on page statements, price and stock on 7 October 2026. CSI publishes no soot, flame or wick data for any of them, so nothing is ranked on performance, and each card says when not to buy.

Skip for now: anything at all during stage 1; a new oil or dye before its arm has failed to pass and the others have been tried; production packs before the passing arm reaches the bottom of the jar; and a non-contact thermometer from us, because we don't list one.

Frequently asked questions