I Want to Use an 8 cm Jar for a Commercial Candle Collection — Should Every Fragrance Be Burn-Tested Separately Even When Fragrance Load Stays at 8%?

Why 8% of one oil is not 8% of another

A fragrance oil is not an inert passenger in the wax. It travels up the wick with the melted wax and burns in the flame. Oils differ in how heavy and viscous they are, how quickly their light top notes evaporate, and how much dense base material they carry: vanillin, resins, woods, oud accords. The same 16 g of each in a 200 g candle is the same weight of very different fuel. Commonly, heavy or vanillin-rich oils leave more carbon on the wick and can make a flame smaller or sootier; light citrus, herbal and aquatic oils often burn more easily and can let the same wick run hotter. These are tendencies, not CSI measurements, and individual oils break them.

That is why a collection built on "one jar, one wax, one wick, one load" still needs a burn per oil. You can fix everything you control; you cannot fix the oil's chemistry, and CSI does not publish per-oil burn data, so there is nothing to look up. Page data can tell you which oils to worry about first (the paper pre-screen in the 12-fragrance odd-ones-out post does exactly that), but it can only set the order of work. It can't clear an oil for sale.

An 8 cm jar makes this sharper. On a 0.35 cm wall (our assumption; measure yours) an 8 cm outside is 7.3 cm inside, a 41.85 cm² pool. At that width the house wick is usually working hard to reach the glass, and at CSI the rung above C2, Cotton Braided, is described by its page as "engineered for large-diameter candles exceeding 8 cm". A wick with little spare capacity is the kind a heavy oil tips into tunnelling and a light oil tips into running hot. The jar itself is yours: CSI lists no 8 cm jar, and its nearest listings are 7 cm and 10 cm.

The policy: a burn-down for every oil, a bracket on a trigger

Collections go wrong in two opposite ways. Makers who test one oil and assume the rest ship a vanilla that soots or a citrus that runs hot. Makers who bracket every oil across every rung spend weeks and hundreds of candles answering questions nobody asked. The two-tier policy separates the questions.

Which direction does the bracket go? The fault tells you. A ring still on the glass after burn 3, with a quiet flame, points up a rung (C2 to Cotton Braided at CSI, which its page describes for "exceeding 8 cm", so at exactly 8 cm it is a border case to test, not assume). A tall flame, soot or mushrooming points down (C2 to C1). A flame near the glass in the lower third points down and is also a stop signal: put the candle out and log it; don't burn it further to "see what happens". Judge heat by what you see, flame height and position, never by touching the glass.

This is deliberately narrower than a tiered screening plan for a very large range; fast early screens for twenty scents are the 20-fragrance tiered burn plan. How many rungs each oil should get when brackets are needed across a ten-scent launch is the sizes-per-fragrance calculator. This page is the floor under both: nobody skips the burn-down.

The minimum file for one fragrance

Burn-downs run side by side, so the file costs bench space more than calendar time: sixteen candles on one clear table, a jar-width apart and away from fans, finish in about the same number of days as two. If your bench holds eight, run the collection in two waves and put the oils you are most worried about in the first.

What skipping costs, family by family

The table ranks the risk of skipping an oil's file by family, using what the pages state (family, notes, vanillin where captured) and common tendencies. It is our judgement, not CSI data, and it exists to set the order of work and the level of attention, never to excuse an oil from tier 1.

Read the "low" rows carefully. Low risk of needing a different wick is not low risk of harm: a light oil that burns easily on a wick chosen for your lead gourmand is exactly the candle that can run hot near the bottom. The file is what catches that, which is why the table has no "skip" column.

Now set the file's cost against what it protects. On a production basis (16 g oil at the 500 g price, 184 g CSI 464 at the 10 kg price, a C2 wick; jars, labour and GST excluded) a 50-candle batch of Vanilla Bean is ₹11,284.08 of materials and a 50-candle batch of Citrus Lemon ₹7,412.08. The Vanilla Bean file costs ₹468.00. A batch is 24 times its file before a single complaint, refund or return.

₹ per fragrance, and the bracket when it is triggered

That is ₹369.44 per fragrance on average in materials used. Every 50 g bottle leaves 18 g after its file: enough for one more candle to settle a split, not enough for a bracket. Cash from nothing is ₹4,042.18 with the C2 50-pack; costs exclude labour, freight, packaging, jars and GST. Note one oddity if you scale up: Blue Ocean is fractionally dearer per gram at larger sizes (₹4.94 at 50 g, ₹4.96 at 500 g and 1 kg), so its 50 g is not a premium at all.

When a trigger fires: the bracket add-on

In the worked example two oils trigger (illustrative, chosen because their families are the likeliest, not because CSI has burn data on them). Vanilla Bean leaves a ring after burn 3 with a quiet flame, so its bracket goes up: three candles on Cotton Braided plus one C2 control. Woody Oud mushrooms and soots, so its bracket goes down: three on C1 plus one C2 control. Each bracket is four candles: 64 g of oil and 736 g of wax.

Tier 2 cash in the example is ₹1,686.80: a second 50 g of each oil (the 18 g left from the file is not enough; each bracket needs 46 g more), another 1 kg and a 500 g bag of CSI 464 (the 56 g left from tier 1 won't cover two brackets; 84 g left after them), a Cotton Braided 10-pack and a C1 20-pack (the C1 10-pack is out of stock on 29 September 2026). Twenty-four candles in all, against 48 if every oil had been bracketed across three rungs at two candles each.

When a passed fragrance goes back on the bench

An approved file stays valid only while everything in its header stays the same. Re-run at least the two-candle file when any of these change: the wax or its supplier (the wax-swap bridging test covers a soy-to-soy switch), the load (changing the load across a collection), the jar or its batch (measure the inside again; a few millimetres move the pool), the wick or its supplier, a heavy dye added to a scent that had none, or a new oil batch that smells or looks different from your retained sample (comparing a new oil batch). A label change that prints a longer session also counts, because the file was burned at the old length.

Diwali is 8 November 2026, 40 days from 29 September 2026. A collection whose files are poured this week, cured about a week and burned in parallel can have most tier-1 verdicts by late October; any oil that triggers will need its bracket and a second full life, and may not make a Diwali shelf. Launch the oils that pass; hold the ones still on the bench. That is the policy working, not failing.

What to buy for the files

Four items: the two that every file shares and two oils that bracket the risk table, chosen on page statements, price and stock on 29 September 2026. CSI publishes no wick-performance data, so nothing is ranked on burn results. Oil candles are counted at 16 g (200 g at 8%) and 12 g (150 g at 8%).

Frequently asked questions