My Two Wicks Are Identical but One Keeps Drowning — Could an Uneven Jar Base or Off-Centre Positioning Cause the Difference?

Why one of two identical wicks drowns

A drowning wick is one whose flame shrinks and goes out while it stands in liquid wax, or one that bends over into the pool. The usual description is that the pool around it has become too deep, or too full of debris, for that wick to keep drawing fuel to a steady flame. In a single-wick candle that points straight at the wick or the pool depth. In a pair, if both wicks are the same size from the same pack and trimmed alike, the useful question is different: what is different about the place where the drowning wick stands?

There are three common answers, and each leaves a measurable trace. The surface tilts. The pool's surface settles level whatever the jar does, so a tilted jar has a deeper pool on its low side; a table that isn't level, or a jar that doesn't sit square on its foot, both do this. The base is uneven. Thick-walled glass can be thicker on one side of the base than the other; a tab on the thick side starts higher and runs out of wick sooner near the bottom. The placement is off. A tab pressed down 5 mm off its mark leaves that wick 18.25 mm from the glass instead of 23.25 mm, 22% closer, and changes how the two flames share the pool.

These are geometry, not burn predictions, and none is proof on its own. A 1° tilt makes only 0.81 mm of difference between two wicks 46.5 mm apart. Against a pool many makers aim to keep about 5–10 mm deep (a commonly cited rule of thumb), 2° is 1.62 mm, about 23% of a 7 mm pool. That rarely drowns a wick alone, but it adds to whatever else is already pushing that wick, and it costs nothing to remove. That is why the audit measures everything and blames nothing until a pattern repeats. The starting geometry for a pair in this jar (spacing equal to the inside radius) comes from the 10 cm cluster hub.

The level-and-position audit

Do the audit on the drowning candle itself and, where it says so, on the empty jars from the same batch. You need a level (a phone level app is fine), a steel ruler, a wooden skewer with millimetre marks, a flat tile or glass sheet, and a marker. Keep hands clear of the flame and never judge the glass by touch: if you check anything while the candle is lit, it is a skewer depth or a photo.

Depth deserves a warning. A deeper pool beside the drowning wick is what you would expect to find, and it can be a result as much as a cause: a pool that is deep for any reason can drown a wick, and a struggling wick changes the pool around it. That is why depth is the last row and is read together with the five before it. In our worked example the drowning wick sits in 14 mm of liquid wax against 9 mm beside its partner, 56% deeper; on a level table with seated tabs and equal positions, that number would send you to the jar check, not to the wick packet.

Checking the jar batch, empty

CSI's Apothecary Glass Candle Jar page gives "Diameter (outer rim): Approx. 10 cm", "Base diameter: Approx. 10 cm" and "Height (without lid): Approx. 10.8 cm", and calls the glass "thick-walled" with a "Curved silhouette with broad base for wax stability". It states no base thickness, no inside diameter and no tolerance, and one set of dimensions covers both sizes. Mass-produced glass varies between batches, so the only way to know your jars is to measure a few. The fuller method for checking a container before you rely on it is in the switching-suppliers post; here is the short version for a drowning pair.

Rock test and rim heights

Stand the empty jar on a flat tile or a sheet of glass. Press down gently on the rim at four points, north, east, south and west: any rock means the foot is not flat. Then measure from the tile to the top of the rim at the same four points. A rim that reads 1.5 mm higher on one side than the other across a 100 mm jar means the jar's own axis leans; for two wicks 46.5 mm apart on that axis, the pool would be 0.70 mm deeper at one of them on a perfectly level table.

Base thickness at both wick positions

Measure the outside height at a point, then drop a ruler inside to the floor at the same point, about 2 cm in from the wall, and subtract. Do it at the two places your tabs sit. Our worked example finds 5 mm on one side and 8 mm on the other: a 3 mm step. On a level table that does not change the pool at first, because the pool surface is level and far above the base. It matters later, because the tab on the thick side sits higher, so that wick reaches its tab sooner; and the floor's heat behaviour near the end is a known source of late uneven burning (see the uneven-bottom post for the single-wick version, and the thick-base post for what a thick base does to the final third).

The rule: when the jar batch is at fault

The audit gives you a row of numbers for each candle. The rule below turns them into a verdict. The thresholds are our judgement, chosen because they can be measured with a ruler and a level; set your own if you prefer, then keep them fixed for the whole test.

The order matters. The surface is removed first because it is free and affects every other reading. Placement comes before the jar because it is under your control and shows up on the very candle you have. The jar batch comes last because it needs repetition: one jar can be odd; a batch is a pattern. The final row is the honest exit. If the surface, jar and placement are all clean and the drowning side moves around, the cause is not where this post looks: debris in the pool, carbon on the wick, or depth late in the life. Those are covered in the late-drowning post.

Changing the wick size is not on the list. A larger pair might stop the drowning, but it makes both flames larger to fix a difference between them, and the partner that was burning correctly may now run too hot. A smaller pair makes drowning more likely. If your audit sends you towards size, it is a separate, controlled test with the jar and placement already fixed; for a 10 cm pair that question belongs to the step-down post and the first-two-sizes post.

Two corrected twins, costed

Pour two twins the same day from one melt: same wax, oil, load and cure as the drowning candle, in two jars that passed the empty-jar check, with tabs seated on stickers, positions marked and measured after the wax sets, and the wick line set across any slope you measured. Burn both on a levelled surface, logging depth beside each wick every session. If neither drowns through the stage where your original did, and depths stay within about 1–2 mm of each other, the suspect was the cause. If one still drowns, the remaining suspects are the wick and the late-life causes.

The worked example matches the cluster hub: the Apothecary jar's 500 gm size at a 500 g fill, 8% of total weight (40 g of oil and 460 g of CSI 464 per candle), on two Eco C2. The fill is our labelled choice; the page says "Fill capacity: ~500 g - 700 g (2 sizes)" without saying whether that is to the brim, so weigh water to your fill line first. With CSI 464, whose page states "Pour Temperature 80°C – 90°C", pour at 80–85°C in this jar, whose page says "Safe for hot-pour up to 85°C": the lower limit governs.

One twin costs ₹641.68: ₹172.00 of oil, ₹233.40 of wax, ₹212.40 of jar, two wicks and two stickers. Costs are worked examples and exclude labour, freight, packaging and GST. From nothing, the cash outlay is ₹1,556.60. If the audit pointed at the table, you need none of it: level the table and burn the candles you have.

What to buy, and what to skip

Five items, in the order the audit and the twins use them, 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 40 g (500 g fill at 8%) and 12 g (the standard 150 g candle).

Frequently asked questions