My Jar Supplier Changed the Internal Diameter from 7.6 cm to 7.9 cm — Is That Difference Large Enough to Require New Wick Testing?

What 3 mm actually does to the candle

The number your jar supplier changed is a diameter, and 7.6 cm to 7.9 cm reads like a rounding difference: 3 mm, or 3.95% wider. The wick does not work on diameter, though. Each burn, it has to liquefy the whole top surface of the candle out to the glass, and that surface is a circle whose area is π × d² ÷ 4. Squaring the diameter is what turns a small-looking change into a bigger one. The old jar's melt-pool area is 45.36 cm²; the new jar's is 49.02 cm². That is 3.65 cm² more wax surface, or +8.05% — roughly double the percentage change you see on the ruler.

The second thing that changes is height. If you keep selling a 200 g candle, the same wax has to spread over a wider floor, so the fill line drops. Solid soy wax is lighter than water — we use a labelled approximate factor of 0.86–0.9 g per ml, which varies by wax, temperature and fill line, so confirm it with a weighed test pour. At the middle of that range, 0.88 g/ml, 200 g of candle takes up 227.3 ml. In the 7.6 cm jar that is a 5.01 cm column of wax; in the 7.9 cm jar it is 4.64 cm, about 3.7 mm lower. Nothing about the formula has changed, but the wick now starts a little deeper below the rim and has a little less depth to burn through.

The last row matters more than it looks. Many makers fill to a line, not to a weight. If your pourer keeps the same fill height in the new jar, each candle quietly becomes 216.1 g, not 200 g — 16.1 g more candle, 1.29 g more oil and 14.8 g more wax at an 8% load. Over a 1,000-candle run that is 14.81 kg of extra wax and 1,288 g of extra oil you did not budget for, and a label weight that is now wrong. Decide which you are holding constant — the weight or the line — before you test, because it changes what you are testing. The wider-and-shorter jar post works the same-weight case in detail, and the 1,000-candle substitution post prices the same-line case.

How big is "big enough"? The sensitivity table

The question in the title — is 3 mm large enough to matter? — has no universal cut-off. CSI does not publish a wick-size-per-diameter chart, and nobody can tell you from a drawing that a particular wick will or will not cope with a particular jar. What geometry can do is show you how fast the load on the wick grows as the diameter creeps up, so you can see where your change sits.

Two readings of that table are worth holding on to. First, the area change is always about twice the diameter change for small changes, so a supplier telling you a jar is "only 4% wider" is telling you the wick has about 8% more to melt. Second, there is no step in the table where the physics suddenly changes. A 1 mm difference (+2.65% area) is within what many makers see between two jars from the same mould run, while a 1 cm difference (+28.05%) is a different candle. Your 3 mm sits between those. That is exactly the zone where a short, controlled comparison is cheaper than an argument.

The retest matrix: what stays valid, what is void

A jar is not part of the formulation, so the tests that describe the formula itself — whether the oil binds in the wax at your load, how the candle cures, its cold throw, its surface appearance after storage — are not voided by a wider jar. The tests that describe how the wick, the wax pool and the glass behave together are voided, because the geometry they measured has changed. The matrix below sorts every test you are likely to have on file into the four words this series uses consistently: KEEP (still valid), CONFIRM (short side-by-side check), REPEAT (full retest) and NEW (a test you never needed before).

Two lines in that matrix deserve a second look. Frosting sits under CONFIRM, not KEEP, even though the wax has not changed, because the glass and the way it cools can differ between jar makers — the only way to know is to look at the test candles after your usual cure. And the full-life burn is conditional: if the upper-jar CONFIRM tests match the control, the late-stage behaviour is unlikely to be the first place a 3 mm difference shows up, but if any early line is worse, repeat the full-life burn before release. If the new jar's base or walls are also different, that condition flips — see the thicker-base post and the thicker-glass post. The broader version of this matrix, for any jar change, lives in the only-the-jar-changed post.

The minimum test: nine candles, two jars, one variable

The protocol below is a working practice, not a standard. It is built to answer one question with as few candles as possible: does the approved wick, in the approved formula, behave the same in the new jar as it did in the old one? Everything except the jar is held constant. The third set of three candles is optional insurance — a bracket wick poured at the same time, so that if the approved wick struggles you already have the answer cured and waiting, instead of starting another week-long cycle.

Consumption rate is the line people skip and later regret. A wider pool can burn a little more wax per hour, and if it does, the burn time on your label is out of date even when every safety line passes. Weigh before and after each burn, divide the loss by the hours, and carry that number into your label. The rest of the sheet is the same one used across this series, so results from the taller-jar post or the frosted-glass post can sit on the same record.

The priced test kit

The kit below is built from listed, in-stock CSI sizes on the 24 September 2026 pull. We use Luxury Soy Wax CSI 464 and Blue Ocean as the worked example; if your approved formula is a different wax or oil, substitute it — the quantities do not change, only the prices. The load is our 8% working assumption, inside CSI 464's stated 10% maximum. Allowance: 5% for pour loss, a labelled working assumption.

The wick choice needs one honest note. In this worked example the approved wick is Eco Thin C1 and the bracket is Eco Thick C2, simply because those are the two Eco sizes CSI lists. We do not publish a size chart that says Thick C2 is right for a 7.9 cm jar or that the two sizes are one step apart for your wax — nobody can say that without a burn. If your approved wick is from another range, bracket it with the next size in that range. The wider-and-shorter post discusses when the bracket is worth pouring and when it is not.

The decision rule: pass, adjust or escalate

CSI publishes no numeric limit for flame height or jar temperature, and we will not invent one. The acceptance test that holds up is relative: the new jar must be no worse than the approved control on every safety line, and within your jar supplier's guidance and the recognised candle fire-safety standard you work to (ASTM F2417 is a commonly cited US one). Performance lines — throw, time to full pool, consumption — get a tolerance you set yourself and write down before you look at the results, so the data cannot talk you into a pass.

The last row is the one people do not expect. Sometimes the test shows that the variation within the new jar lot is larger than the change you were worried about — one candle reaches the wall in two hours, its neighbour in three. That is not a wick problem. It is a measurement problem, and the answer is to measure more jars and talk to the supplier, not to resize the wick for the widest jar in the box. The alternative-container post includes a scoring rule for exactly that kind of lot comparison.

Release and record

A pass is not finished until it is written down, because the next time someone asks why the Blue Ocean candle is in a different jar, the answer has to be on paper, not in your memory. The record does not need software. A single row in a spreadsheet or a page in a bound notebook is enough if it holds the lines below.

Keep one unburned candle from each set as a retained sample, dated and labelled. If a customer complaint arrives in three months, that candle is the only way to show whether the problem is the jar or something that changed later. The full version of this record — with a form you can copy — is the simple change-control form post, and the reproducibility record post covers how to write jar dimensions so they mean the same thing a year from now.

Which jar change are you making?

This page is the hub for the jar-change questions in this series. Each sibling takes one kind of container change and works its own geometry, its own retest matrix and its own test. If your change is not simply "a bit wider", go to the one that matches.

Two posts outside this cluster sit alongside it. If you change only the jar is the general retest list for any container change, in the change-control part of the series, and the revised-formulation post draws the line between a change you confirm and a change so large the candle should be treated as a new product.

Where jar changes go wrong

The first mistake is testing the wrong jar. A supplier sends three samples, they are the best-finished jars of the run, and the production lot measures differently. Test with jars from the actual lot you will use. The second is changing something else at the same time — a new jar and a new wick lot and a new wax delivery in one week. If the candle then fails, you cannot tell which change did it, and the two-changes-at-once post is the long way back. Change the jar first, alone.

The third is calling it done after one burn. A wider jar often looks fine on the first burn, when the wick has a fresh, flat top and a cool jar. The differences show up by the third or fourth cycle, when the tunnel ring either closes or grows. Burn at least the early cycles on all six test candles before deciding. The fourth is forgetting the label. If the new jar burns a little more wax per hour, the burn time you print is no longer true, and a label that promises more hours than the candle gives is a complaint you wrote yourself.

Finally, do not let this single test harden into a rule of thumb. "3 mm didn't matter last time" is true for that wick, that wax, that oil and that jar. The next change — a different oil in the same jar, or the same oil in a jar with a thicker base — is a new question. The geometry takes two minutes; do it every time.

Frequently asked questions